{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7c9e2048",
   "metadata": {},
   "source": [
    "# Fine-Tuning ChatGPT for Different Data Sets"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e8ddc61",
   "metadata": {},
   "source": [
    "# Imports\n",
    "Here we import all the necessary packages and modules for our fine-tuning process.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b410a579",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "e:\\Dropbox\\projects_active\\eth_uzh_hatespeech_llm_finetuning\\src\\gpt_finetuning\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import glob\n",
    "import pprint\n",
    "import time\n",
    "import argparse\n",
    "from ast import literal_eval\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from tqdm import tqdm\n",
    "\n",
    "import wandb\n",
    "from openai import OpenAI\n",
    "import pandas as pd\n",
    "\n",
    "from utils_2 import (\n",
    "    dataset_has_format_errors,\n",
    "    check_token_statistics_and_cost_estimate,\n",
    "    load_dataset_task_prompt_mappings,\n",
    "    write_jsonl,\n",
    ")\n",
    "from utils_1 import task_num_to_task_name, dataset_num_to_dataset_name, plot_count_and_normalized_confusion_matrix, task_to_display_labels\n",
    "\n",
    "# Get the notebook's full path\n",
    "notebook_path = os.getcwd()\n",
    "\n",
    "# Set module_dir to the notebook's directory\n",
    "module_dir = notebook_path\n",
    "\n",
    "#api_key_project = \"\" #MK's Key\n",
    "api_key_project = \"\" #new PRODIGI key\n",
    "\n",
    "client = OpenAI(\n",
    "    api_key=api_key_project,  # This is the default and can be omitted\n",
    ")\n",
    "\n",
    "\n",
    "WANDB_PROJECT_NAME = \"chatgpt_annotations_llm_comparison\"\n",
    "MODEL_NAME = 'gpt-4o-mini-2024-07-18'\n",
    "COMPLETION_RETRIES = 50\n",
    "\n",
    "print(notebook_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "772bcc76",
   "metadata": {},
   "source": [
    "# Utility Functions\n",
    "These functions provide various utilities for tasks such as data loading, token statistics, and other preprocessing steps.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa7e1cb4",
   "metadata": {},
   "source": [
    "## Configuration Values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c79142f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Configuration Variables\n",
    "# Manually set the variables that were previously handled by argparse\n",
    "\n",
    "# Type of task to run inference on\n",
    "task = 1  # Choices: [1]\n",
    "\n",
    "# Dataset to run inference on\n",
    "dataset = 4  # Choices: [1, 2, 3, 4, 5, 6]\n",
    "\n",
    "# Expert Dataset for Second Evaluation\n",
    "dataset_eval = 'x' # Choises: ['x']\n",
    "\n",
    "# Size of the sample to generate\n",
    "sample_size = '250'  # Choices: ['100','200']\n",
    "\n",
    "# Path to the directory to store the generated samples\n",
    "output_dir = '../../annotations/chatGPT/output_fine_tuning/'\n",
    "\n",
    "# Random seed to use\n",
    "seed = 2019\n",
    "\n",
    "# Path to the directory containing the datasets\n",
    "data_dir = '../../annotations/chatGPT/input_fine_tuning/'\n",
    "\n",
    "# Whether to use the full label\n",
    "not_use_full_labels = False\n",
    "\n",
    "# Path to the dataset-task mappings file\n",
    "dataset_task_mappings_fp = os.path.normpath(os.path.join(module_dir, '../../task_mappings', 'dataset_task_mappings.csv'))\n",
    "\n",
    "# Whether to rewrite the dataframe in OpenAI format\n",
    "rewrite_df_in_openai = True\n",
    "\n",
    "# Number of epochs to train the model\n",
    "n_epochs = 5\n",
    "\n",
    "# Batch Size for learning\n",
    "n_batch = 20\n",
    "\n",
    "# Name of the run\n",
    "run_name = 'finetune_chetGPT_hatespeech_citi'\n",
    "\n",
    "# Temperature to use when generating text\n",
    "temp = 0.0\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71e81ba8",
   "metadata": {},
   "source": [
    "## Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4ef6fa17",
   "metadata": {},
   "outputs": [],
   "source": [
    "def map_label_to_completion(label: str, task_num: int, full_label: bool = True) -> str:\n",
    "    new_label = ''\n",
    "\n",
    "    if task_num == 1:\n",
    "        if full_label:\n",
    "            if str(label) in ['1.0', '1']:\n",
    "                new_label = 'HATE SPEECH'\n",
    "            elif str(label) in ['2.0', '2']:\n",
    "                new_label = 'TOXIC SPEECH'\n",
    "            else:\n",
    "                new_label = 'KEINE HATE SPEECH'\n",
    "            assert new_label in ['HATE SPEECH', 'TOXIC SPEECH', 'KEINE HATE SPEECH']\n",
    "        else:\n",
    "            if str(label) in ['1.0', '1']:\n",
    "                new_label = 'A'\n",
    "            elif str(label) in ['2.0', '2']:\n",
    "                new_label = 'C'\n",
    "            else:\n",
    "                new_label = 'B'\n",
    "            assert new_label in ['A', 'B', 'C']\n",
    "\n",
    "    return new_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "02978fd4",
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_train_and_eval_sets(data_dir: str, dataset_num: int, task_num: int, sample_size: int, dataset_eval:str) \\\n",
    "        -> dict[str, pd.DataFrame]:\n",
    "    datasets = dict()\n",
    "\n",
    "    train_dataset_task_files = glob.glob(os.path.join(data_dir, f'ds_{dataset_num}__task_{task_num}_train_set*.csv'))\n",
    "    eval_set_name = f'ds_{dataset_num}__task_{task_num}_eval_set'\n",
    "    datasets[eval_set_name] = pd.read_csv(os.path.join(data_dir, eval_set_name + '.csv'), encoding='utf-8')\n",
    "\n",
    "    # Load the additional evaluation dataset specified by dataset_eval\n",
    "    second_eval_set_name = f'ds_{dataset_eval}__task_{task_num}_full_eval'\n",
    "    datasets[second_eval_set_name] = pd.read_csv(os.path.join(data_dir, second_eval_set_name + '.csv'), encoding='utf-8')\n",
    "\n",
    "    if sample_size == 'all':\n",
    "        train_dfs_ = {fn.strip('.csv'): pd.read_csv(fn, encoding='utf-8') for fn in train_dataset_task_files}\n",
    "        datasets.update(train_dfs_)\n",
    "    else:\n",
    "        train_df_fn = f'ds_{dataset_num}__task_{task_num}_train_set_{sample_size}'\n",
    "        datasets[train_df_fn] = pd.read_csv(os.path.join(data_dir, train_df_fn + '.csv'), encoding='utf-8')\n",
    "\n",
    "        if train_df_fn not in [os.path.basename(fn).strip('.csv') for fn in train_dataset_task_files]:\n",
    "            raise ValueError(f\"Sample size {sample_size} not found for\"\n",
    "                             f\" dataset {dataset_num} and task {task_num}\")\n",
    "\n",
    "    return datasets\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "904e16c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_training_example(system_prompt, user_prompt_format, user_prompt_text, completion):\n",
    "    return {'messages': [\n",
    "        {'role': 'system',\n",
    "         'content': system_prompt},\n",
    "\n",
    "        {'role': 'user',\n",
    "         'content': user_prompt_format.format(text=user_prompt_text)},\n",
    "\n",
    "        {'role': 'assistant',\n",
    "         'content': completion}\n",
    "    ]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "bdf692a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def upload_datasets_to_openai(output_dir, not_use_full_labels, rewrite_df_in_openai, datasets):\n",
    "    hatespeech_open_ai_metadata = list()\n",
    "\n",
    "    df_id_metadata = pd.DataFrame() if not os.path.exists('hatespeech_open_ai_metadata.csv') \\\n",
    "        else pd.read_csv('hatespeech_open_ai_metadata.csv')\n",
    "\n",
    "    for df_name, df in datasets.items():\n",
    "        df_jsonl_filename = os.path.join(output_dir, 'temp', df_name + '.jsonl')\n",
    "        write_jsonl(data_list=df['openai_instance_format'].tolist(), filename=df_jsonl_filename)\n",
    "\n",
    "        if not_use_full_labels:\n",
    "            df_name += '_single_letter_labels'\n",
    "\n",
    "        if (not rewrite_df_in_openai and\n",
    "                (len(df_id_metadata) > 0 and df_name in df_id_metadata['df_name'].tolist())):\n",
    "            print(f\"Dataset {df_name} already uploaded to OpenAI\")\n",
    "            continue\n",
    "\n",
    "        print(f\"Uploading {df_name} to OpenAI\")\n",
    "        df_response = client.files.create(\n",
    "            file=open(df_jsonl_filename, \"rb\"), purpose=\"fine-tune\"\n",
    "        )\n",
    "        df_response_dict = df_response.to_dict()\n",
    "        df_file_id = df_response_dict[\"id\"]\n",
    "\n",
    "        # Wait until the file is processed\n",
    "        while True:\n",
    "            file = client.files.retrieve(df_file_id)\n",
    "            file_dict = file.to_dict()\n",
    "            if file_dict[\"status\"] == \"processed\":\n",
    "                break\n",
    "            time.sleep(15)\n",
    "        hatespeech_open_ai_metadata.append({'df_name': df_name, 'file_id': df_file_id})\n",
    "\n",
    "    df_id_metadata = pd.concat([df_id_metadata, pd.DataFrame(hatespeech_open_ai_metadata)])\n",
    "    df_id_metadata.to_csv('hatespeech_open_ai_metadata.csv', index=False)\n",
    "\n",
    "    return df_id_metadata\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ace921b9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def fine_tune_chat_gpt(evaluation_file_id, training_file_id, model_name, n_epochs):\n",
    "    response = client.fine_tuning.jobs.create(\n",
    "        training_file=training_file_id,\n",
    "        validation_file=evaluation_file_id,\n",
    "        model=\"gpt-4o-mini-2024-07-18\",\n",
    "        suffix=model_name,\n",
    "        hyperparameters={\"n_epochs\": n_epochs,\n",
    "                         \"batch_size\": n_batch}\n",
    "    )\n",
    "\n",
    "    response_dict = response.to_dict()\n",
    "    job_id = response_dict[\"id\"]\n",
    "    print(\"Job ID:\", response_dict[\"id\"])\n",
    "    print(\"Status:\", response_dict[\"status\"])\n",
    "\n",
    "    # Wait until the job is done\n",
    "    while True:\n",
    "        job = client.fine_tuning.jobs.retrieve(job_id)\n",
    "        job_dict = job.to_dict()\n",
    "        if job_dict[\"status\"] == \"succeeded\":\n",
    "            break\n",
    "        elif job_dict[\"status\"] == \"failed\":\n",
    "            raise Exception(\"Training failed: %s\" % job_dict[\"error\"])\n",
    "        time.sleep(30)\n",
    "\n",
    "    return job_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2026f811",
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_and_log_finetuning_event_history(job_id):\n",
    "    response = client.fine_tuning.jobs.list_events(fine_tuning_job_id=job_id)\n",
    "    response_dict = response.to_dict()\n",
    "    events = response_dict[\"data\"]\n",
    "    events.reverse()\n",
    "    for event in events:\n",
    "        print(event[\"message\"])\n",
    "\n",
    "    # Log events\n",
    "    for event in events:\n",
    "        if event['type'] != 'metrics':\n",
    "            continue\n",
    "        data = event['data']\n",
    "        wandb.log(\n",
    "            {\n",
    "                \"train_loss\": data.get(\"train_loss\"),\n",
    "                \"valid_loss\": data.get(\"valid_loss\"),\n",
    "                \"train_mean_token_accuracy\": data.get(\"train_mean_token_accuracy\"),\n",
    "                \"valid_mean_token_accuracy\": data.get(\"valid_mean_token_accuracy\")\n",
    "            },\n",
    "            step=data.get('step', 0)\n",
    "        )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ebc50a97",
   "metadata": {},
   "source": [
    "# Main Implementation\n",
    "This section contains the main code for fine-tuning the ChatGPT model. It includes setup, data preparation, training, and evaluation steps.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed1b7246",
   "metadata": {},
   "source": [
    "### Connect to WandB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "37564c84",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Failed to detect the name of this notebook, you can set it manually with the WANDB_NOTEBOOK_NAME environment variable to enable code saving.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[34m\u001b[1mwandb\u001b[0m: Currently logged in as: \u001b[33mkublima\u001b[0m (\u001b[33mdigdemlab\u001b[0m). Use \u001b[1m`wandb login --relogin`\u001b[0m to force relogin\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "wandb version 0.19.7 is available!  To upgrade, please run:\n",
       " $ pip install wandb --upgrade"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Tracking run with wandb version 0.15.11"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Run data is saved locally in <code>e:\\Dropbox\\projects_active\\eth_uzh_hatespeech_llm_finetuning\\src\\gpt_finetuning\\wandb\\run-20250228_105559-avmc8new</code>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Syncing run <strong><a href='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new' target=\"_blank\">finetune_chetGPT_hatespeech_citi</a></strong> to <a href='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison' target=\"_blank\">Weights & Biases</a> (<a href='https://wandb.me/run' target=\"_blank\">docs</a>)<br/>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       " View project at <a href='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison' target=\"_blank\">https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison</a>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       " View run at <a href='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new' target=\"_blank\">https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new</a>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<button onClick=\"this.nextSibling.style.display='block';this.style.display='none';\">Display W&B run</button><iframe src='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new?jupyter=true' style='border:none;width:100%;height:420px;display:none;'></iframe>"
      ],
      "text/plain": [
       "<wandb.sdk.wandb_run.Run at 0x1f0e07377c0>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Section 1/2\n",
    "# Initialize the Weights and Biases run\n",
    "wandb.init(\n",
    "    # set the wandb project where this run will be logged\n",
    "    project=WANDB_PROJECT_NAME,\n",
    "    name=run_name if run_name != '' else f'{MODEL_NAME}_ds_{dataset}_task_{int(task)}'\n",
    "                                                    f'_sample_{sample_size}_epochs_{n_epochs}'\n",
    "                                                    f'_full_label_names_{str(not not_use_full_labels)}'\n",
    "                                                    f'_temp_{temp}',\n",
    "\n",
    "    # track hyperparameters and run metadata\n",
    "    config = {\n",
    "        \"model\": MODEL_NAME,\n",
    "        \"dataset\": dataset_num_to_dataset_name[int(dataset)],\n",
    "        \"task\": task_num_to_task_name[int(task)],\n",
    "        \"epochs\": n_epochs,\n",
    "        \"temp\": temp\n",
    "    }\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "58f1df25",
   "metadata": {},
   "source": [
    "### Load and Process Data Sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "2454b3ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 3\n",
    "# Load the dataset and filename\n",
    "dataset_idx, dataset_task_mappings = load_dataset_task_prompt_mappings(\n",
    "    dataset_num=dataset, task_num=task, dataset_task_mappings_fp=dataset_task_mappings_fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d35437fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 5\n",
    "# Load the train and eval datasets\n",
    "datasets = load_train_and_eval_sets(\n",
    "    data_dir=data_dir, dataset_num=dataset, task_num=task, sample_size=sample_size, dataset_eval=dataset_eval)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "9f379906",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Artifact prompts>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Section 4\n",
    "# Get information specific to the dataset\n",
    "label_column = dataset_task_mappings.loc[dataset_idx, \"label_column\"]\n",
    "system_prompt = dataset_task_mappings.loc[dataset_idx, 'zero_shot_prompt']\n",
    "user_prompt_format = dataset_task_mappings.loc[dataset_idx, 'user_prompt']\n",
    "    \n",
    "# Log the system prompt and user_prompt_format as files in wandb\n",
    "prompts_artifact = wandb.Artifact('prompts', type='prompts')\n",
    "with prompts_artifact.new_file('system_prompt.txt', mode='w', encoding='utf-8') as f:\n",
    "    f.write(system_prompt)\n",
    "with prompts_artifact.new_file('user_prompt_format.txt', mode='w', encoding='utf-8') as f:\n",
    "    f.write(user_prompt_format)\n",
    "wandb.run.log_artifact(prompts_artifact)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "47917349",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 6\n",
    "# Generate the training and evaluation examples in the way expected by the Open AI API to finetune chatgpt3.5\n",
    "preprocessed_output_dir = os.path.join(\n",
    "    output_dir, 'preprocessed', 'full_name_labels' if not not_use_full_labels else 'single_letter_labels')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "5a9616d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 7\n",
    "for df_name, df in datasets.items():\n",
    "    df['completion_label'] = df[label_column].map(\n",
    "        lambda label: map_label_to_completion(label=label, task_num = task,\n",
    "                                              full_label=not not_use_full_labels)\n",
    "        )\n",
    "    df['openai_instance_format'] = df.apply(\n",
    "        lambda row: create_training_example(\n",
    "            system_prompt=system_prompt, user_prompt_format=user_prompt_format,\n",
    "            user_prompt_text=row['text'],\n",
    "            completion=row['completion_label']\n",
    "        ),\n",
    "        axis=1\n",
    "    )\n",
    "    df['openai_instance_without_completion'] = df['openai_instance_format'].map(lambda x: x['messages'][:-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8bcd8c12",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Check for errors ds_4__task_1_train_set_250 set: \n",
      "No errors found\n"
     ]
    }
   ],
   "source": [
    "# Section 8\n",
    "print(f'Check for errors {df_name} set: ')\n",
    "assert not dataset_has_format_errors(df['openai_instance_format'].tolist()), f\"Errors found in {df_name}\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "9b1d7833",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 9\n",
    "df.to_csv(os.path.join(preprocessed_output_dir, df_name + '.csv'), index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "24cdc6b4",
   "metadata": {},
   "source": [
    "### Upload Training Dataset to OPEN AI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2fd9b187",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploading ds_4__task_1_eval_set to OpenAI\n",
      "Uploading ds_x__task_1_full_eval to OpenAI\n",
      "Uploading ds_4__task_1_train_set_250 to OpenAI\n"
     ]
    }
   ],
   "source": [
    "# Section 10\n",
    "# Create jsonl file and upload to OpenAI\n",
    "df_id_metadata =upload_datasets_to_openai(output_dir, not_use_full_labels, rewrite_df_in_openai, datasets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "705758da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Section 11\n",
    "# delete all files in the temp folder\n",
    "os.system(f\"rm -rf {os.path.join(output_dir, 'temp')}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4519ab5a",
   "metadata": {},
   "source": [
    "### Finetune chatGPT with the Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "b0ec6e82",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 12\n",
    "# Run training on the train_df samples selected\n",
    "eval_set_name = f'ds_{dataset}__task_{task}_eval_set'\n",
    "eval_df = datasets[eval_set_name]\n",
    "for df_name, df in datasets.items():\n",
    "    if df_name == eval_set_name:\n",
    "        continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a76b1ab7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finetuning ds_4__task_1_train_set_250\n",
      "--------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "# Section 13\n",
    "print(f\"Finetuning {df_name}\")\n",
    "print('-' * 50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "a1036bc7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 14\n",
    "# Log in wandb.config the dataset sample size used\n",
    "sample_size = df_name.split('_')[-1]\n",
    "wandb.config['trainset_size'] = sample_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d0860e65",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Statistics for the evaluation set: \n",
      "\n",
      "#### Distribution of num_messages_per_example:\n",
      "min / max: 3, 3\n",
      "mean / median: 3.0, 3.0\n",
      "p5 / p95: 3.0, 3.0\n",
      "\n",
      "#### Distribution of num_total_tokens_per_example:\n",
      "min / max: 847, 999\n",
      "mean / median: 879.8375634517766, 869.0\n",
      "p5 / p95: 853.0, 920.0\n",
      "\n",
      "#### Distribution of num_assistant_tokens_per_example:\n",
      "min / max: 4, 6\n",
      "mean / median: 5.241116751269035, 6.0\n",
      "p5 / p95: 4.0, 6.0\n",
      "\n",
      "0 examples may be over the 4096 token limit, they will be truncated during fine-tuning\n",
      "Dataset has ~346656 tokens that will be charged for during training\n",
      "By default, you'll train for 5 epochs on this dataset\n",
      "By default, you'll be charged for ~1733280 tokens\n"
     ]
    }
   ],
   "source": [
    "# Section 15\n",
    "# Check statistics and possible cost for each dataset\n",
    "print('Statistics for the evaluation set: ')\n",
    "eval_tokens = check_token_statistics_and_cost_estimate(\n",
    "    datasets[eval_set_name]['openai_instance_format'].tolist(), target_epochs=n_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "254bdfe1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Statistics for the training set: \n",
      "\n",
      "#### Distribution of num_messages_per_example:\n",
      "min / max: 3, 3\n",
      "mean / median: 3.0, 3.0\n",
      "p5 / p95: 3.0, 3.0\n",
      "\n",
      "#### Distribution of num_total_tokens_per_example:\n",
      "min / max: 847, 997\n",
      "mean / median: 880.952, 868.5\n",
      "p5 / p95: 853.0, 928.2\n",
      "\n",
      "#### Distribution of num_assistant_tokens_per_example:\n",
      "min / max: 4, 6\n",
      "mean / median: 5.072, 5.0\n",
      "p5 / p95: 4.0, 6.0\n",
      "\n",
      "0 examples may be over the 4096 token limit, they will be truncated during fine-tuning\n",
      "Dataset has ~220238 tokens that will be charged for during training\n",
      "By default, you'll train for 5 epochs on this dataset\n",
      "By default, you'll be charged for ~1101190 tokens\n"
     ]
    }
   ],
   "source": [
    "# Section 16\n",
    "print(f'Statistics for the training set: ')\n",
    "train_tokens = check_token_statistics_and_cost_estimate(df['openai_instance_format'].tolist(),\n",
    "                                                        target_epochs=n_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "91aed7d5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of tokens: 566894\n",
      "\n",
      "#### Estimated cost: 4.54 USD\n"
     ]
    }
   ],
   "source": [
    "# Section 17\n",
    "print(f\"Total number of tokens: {eval_tokens + train_tokens}\\n\")\n",
    "print(f\"#### Estimated cost: {round((eval_tokens + train_tokens) * (0.008 / 1000), 2)} USD\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "5a4cb776",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 18\n",
    "# Run the fine-tuning job\n",
    "training_file_id = df_id_metadata.loc[df_id_metadata['df_name'] == df_name, 'file_id'].values[0]\n",
    "evaluation_file_id = df_id_metadata.loc[df_id_metadata['df_name'] == eval_set_name, 'file_id'].values[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "dd12c487",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ID file-7uuc1F6fvyEwT1xLBRNge6, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-KdF5re7d3Ktp78nDoih4cv, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-KVZSuPeQ5bm1hUeB3HbBda, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-YTHYeCaqc1FAbJzzeohqVj, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-CkNsjwqW5dHBkGwHkwk2MF, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Hj54dEWPfxBFvuxBtPfEXN, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Qucqe1Wf1ncWJh2QuQmQQB, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-8KMCCUopGaBv9sb9hu873T, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-MrYSpKPehkfgDrt8KQrsQH, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-YHDeZA3WKQzWmdM187ibrY, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-FNLACfbPrqhQsafeqh9Cxu, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-L2qvkYtLxkLrWXbmpbBffn, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-GMNfA5ixtoiwSrCT1UeAjs, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Wcji2TWS26x8C4opyNP7XA, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-AkV1F8S2x7kTCrug3YyWfr, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-UgfhS8NZQZS5hc6H8yMtnp, Filename: ds_4__task_1_train_set_250.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-X4TFb4Chb6yYYzpHWevgaM, Filename: ds_x__task_1_full_eval.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-QMc4ct91cYA64cy7Y1AqcD, Filename: ds_4__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-NVpetMXnBFsEsjMyjQxSNL, Filename: framing_1224_ds_3__task_2_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-1yr5a4KhFJqUuufvVtHqs4, Filename: framing_1224_ds_3__task_1_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-CQQGRMeP6fkYrEbbhzdSxw, Filename: framing_1224_ds_2__task_1_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-NmQECGAFut1AiVRnU7FmnK, Filename: framing_1224_ds_2__task_2_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-PHBg2poT9xURaJ3ExPzwad, Filename: framing_1224_ds_1__task_2_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-EyNpfh862Ty25gmGhWkkxG, Filename: framing_1224_ds_1__task_1_full__for_zero_shot_classification.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-GSf8qaPHoaDoamCveimbNiFB, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-SujH6TJ97As20y4IZ8RtRyiV, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-aQts8mp4xCAdTGUPBEOV2GgI, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-ntXye2twvoOiVvJgEAUd7JDl, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-zFIGpRdmbIISFXQ7e69frerz, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-7luQQ41rohn1xBCVowqZKAXi, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-KDrtDdZe9qEPctHOhQE0Un6X, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-7vh1nX8GCP2ljbC7w1ebmIyy, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-SPcSROnitZkcsjQY14Rt1a1W, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Fbf1GpdeECKEnW5aClivtGca, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Hk0sKyC0jFEyCnQvvovs9w7E, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-sRdOlgp0gbvebtr7Wh3XHP0W, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-BFcApSsiRmSVQYt0HbNDsihA, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-WA7bv3CdFcv0uGPe0XZBbr1f, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Ngbcu44ZIdCqlSBt4cX7BlIg, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-IFYiuJcdGjAZ6fskQk6HFVE7, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-z1yw0ZTlVWPhBDF03HPmuBG9, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-8bsVydvCPcFBxCCJUAHPBS4E, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Szpl2ZaePH7ObwQESrcWuPyD, Filename: chatGPT_template_1_ds_1__task_1_train_set_50.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6MOHjmrQJ5pxByRr1GTDbkAx, Filename: chatGPT_template_1_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-UJjCsOfubevzkCfhB7IjDGF2, Filename: chatGPT_template_1_ds_1__task_1_train_set_50.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6YOjqBpfMIwdGxJSFOR8APCm, Filename: chatGPT_template_1_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-aaBIdmuhhrdFqwWAMg2CHpd9, Filename: chatGPT_template_1_ds_1__task_1_train_set_50.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-fBpYTt0iyp7BzCoTsZlCqTmX, Filename: chatGPT_template_1_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-me4dQj6rGtoLxhX4DDp0XI3x, Filename: chatGPT_template_1_ds_1__task_1_train_set_50.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-ysRxeLlGnBBvnLWTHVTJzCuT, Filename: chatGPT_template_1_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-zAla8TSTijw8gtZ4AL5HnvV9, Filename: chatGPT_template_1_ds_1__task_2_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-5dcIZI1WOtK3YdgQ6Dxw8pAf, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-xeT0iDrMZK30xkq72rwK5ViX, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-hOAZdrgK5TpiQx8SAEZO4dwB, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-QoDnX4qHxcD4n4VX0XDzFtyx, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Ix3XWXtLObmM2SOZppeJdI0h, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-JolFGtlcz3vYPQKYdGnyoU8V, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-32xLT3SUhtkly5noVEo29ESa, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-oIheloAMSvVbkSOKfWRNIXwB, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-kKiKEYHcg73q5cQg1LIn2eEl, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-IUSSV34mGvDoascOGlkRcqww, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-hl3HtE8b4IkDstmzR6IRXf9x, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-4mi7XGMLOmfiYQdgwAQl7D3a, Filename: chatGPT_template_1_ds_1__task_1_eval_set_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-JzSAXJwsjrJ7uJmKpBrGOO8Z, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-3vSRfUe8RYfxcx097CjaZmmN, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-7Exoum0vYukjZfchBboH0xtj, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-moyEaJmfDUhhl3QiKfFx6H3P, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-96YZnOEyrTVffag9X3RDeQ4y, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-9KzuJVJBtiaUtxkmaT4OK14I, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-q2OJ0Q0nfcKp0XWQQraVCZNk, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-5NickCCt2e85bi433iztDptu, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Tn8WViAoCXebWExlmbJei3Y6, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-q5FoOQwbmGqNKBLtDPqXn8U0, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-ZTViymT8QcqFaIyX726Y6ptK, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-k9BOcYGg02vRv9rshHy32THQ, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-8cTVW4jy9Z8JVdHqJN0guZWe, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-n2v7Wj6Bw9Jaol8U3QU70KU4, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-NkqkMqTEzWSzL9Onjferm0vC, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-3J9mJNvOrnRFXraUqF0WXF8p, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-QKVZWd7O8hoDSSu19Jx8cjjT, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-CLvyIf7rl341jefbsKNkbVqN, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-EPWzGdHpcfE5DBk3pSFjAFv6, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-GE2qoBVwnDKs9EqmhS8vUGHN, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-GBZdHM0rFhYZ1JFMWQE4yMwC, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-YOlvlNnJBKu2n3MGiXBhQK1D, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-VbpYzMkauOm9fjeg8Ahoesva, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-r4WOisMaM2z7B8Q962elRwqi, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Dp7BpSF6MTO9AxB3quGakhms, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-aEj80Tub1xC9quybHt0bP6mi, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-95ldOzQUyhAHbvxZwCGVoJJw, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-XLTESwGI2ls284sFR4GuiBwx, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-WAktOrFUgU97O7AuqbLrVqjq, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-ozbG9U6DMpHvASRqeI1bNwAh, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-kDKJaLe7XtnOSIEB1ggw1VUT, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-0iLjf3CCrURULXH1o5jubWmk, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-UZWhWCAGYCZs6LHRpMj2KF1k, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-IObeEBRHQtWiICc24JeQv50c, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-YqIpmBwPcGP4lLGaeIHEv9a3, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-COeyGjABljcpmRc6cdVz17Gz, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-J51D8JJ2gp2DpckU50oHbkAr, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-4bWLQXxnBjImvZyIUnVdOBEi, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-juILSyJJ7WpkucBfaIcXyIfu, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-LfmF7UnkoZyca5i2mRadD8Mc, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-LrGYgwk8QvmuAzE7BH7Xuytd, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-eb7VEJA1go1U95svgTrksQic, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-S3tbM8OGyu8dFInAyKAU4rrO, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-6TNEEXZI1lmjvsWwsrZyEEmn, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Ogn48odBcYvKdKvT57lJKJ1r, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-oajTmvDdTXky7PMKx8wEPWLb, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-nUTGBOXdb4V0eCeIX4L3ukfg, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-4z76A7P62sZz883x9XYIYtuf, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-S2AmMcCiOMNb3TmcSRckRDuh, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-YoU8VImUQM7g7KXs3QUNnUAE, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-S8gf0B3YyvNsLM1sjKsLnjFG, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-J3fW3wGnApkJvZMowWDkSl59, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-ZMrdUjSozBmtSDNhtF1K9ZCR, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-5Sh7LrMryRXcD2sQXTnbCZXt, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-l8R4tLmQBIuWAHBJAxL7NLMK, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-C00D7zhvcVYKRZ0OgNjBLlus, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-2TRNpLfhPyZhkP2sLAKKVwzs, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-OIyCZehgLdCrOhzcygWfnQBb, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-AvHNF4wDozBiTuFmAgVe06nk, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-4DrxxwI3LDHaBh0dDJ5aU0o8, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-c2htPty3HFocKjw9iknjX6H3, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-SVq8GARbAPkmNvPpROKggpcR, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-DCNCq6txg4zjAtQceN9aLSlu, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-sSmOG4acaDBXczwcw81i8Qpa, Filename: finetune_val_binary_new_18.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-11qJEE5wXTcoFsMXh9On1f9E, Filename: finetune_train_binary_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-VMYH6yDeeiK0oYBzCK0eeeOn, Filename: finetune_train_binary_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-t1tKPI0fYGquHnO5WG8NudEp, Filename: finetune_train_binary_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-VZHeksdqqnLs0rTEcda2gSgN, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-iy5QTWPn2axe8vjzbF4LpXJQ, Filename: finetune_val_binary_self_inst_60.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-RwlQGLZEko5MpVE0ESIc8LG6, Filename: finetune_train_binary_new_with_initial.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-05JXpB320ehpTTlmdFLSOtiv, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-lqHCGTQfbyecp4l2BZyprdkJ, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Qb7sHQ8ZpRgTH4W5MBUbDFL1, Filename: finetune_val_binary_self_inst_60.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-sd5UcMG8pBZzujhz01baS1Nb, Filename: finetune_train_binary_new_sub_18.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-XhGWQM7bQNBwf1EyeS2Ix2Ck, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-6pds20kwacIRjH35Fp6Z9CYL, Filename: finetune_val_binary_new_sub_18.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-lCp9sI47LIGR5iTb4HTPfRcS, Filename: finetune_train_binary_self_inst_60.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-5Gs3MdDOeZmjQm6xXhPHJ1J2, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-IPYo5kgNEDoi5t3GB6XgZMAc, Filename: finetune_validation_bi_subset_newSi.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-snP4jgtnnZskU4o35Kyx28Ly, Filename: finetune_training_bi_subset_newandSi.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-9Pk99KEW4kkpxlqpD0IucVXq, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-IiHaRgIqvP98g6lwJjJW5YsJ, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-4R4Xa4mRS1OwjrY5yjRxjanP, Filename: finetune_validation_bi_new_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6XOK49zMPDGgjDKGvoPPyfMI, Filename: finetune_training_bi_SI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-ZIpSWa1V0pZmrsHt0PdTcP1y, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Wbjyx6YhNQZvEYrNywdroHq8, Filename: finetune_validation_bi_subset_newSi.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-OxaA0EIqVW61TwOxQfoHDF15, Filename: finetune_training_bi_subset_newandSi.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-imTBkokJhPyhAN1tWRj3I8Q5, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-NdOAVPd3iBKV5xwkXFAUPCrt, Filename: finetune_validation_bi_subset_newSi.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-yLZ8qErKAwLJlSIwK4ToSS02, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-TbXlvT6dbXAAFUC0JE9yp8p0, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-c9bc7ZhVUI4IrLaf3oRpseQP, Filename: finetune_validation_bi_new_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-cgp5RnKalgG27p0AuI36gFgN, Filename: finetune_training_bi_SI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-baSfbflfBnQPBwrEkOkzEz5X, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-ll7j5mvyZIuXuCfUmwJzey9f, Filename: finetune_validation_bi_SelfI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-8eLJDE77wC5qQNsxW64SW5Lf, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Oozzlzq8PDhj4350DLmiGOu7, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-6q3Ike63AdAEFWeb1wgtTmFN, Filename: finetune_validation_bi_SelfI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-kxUyxwFXYFCaNzhjWwdUvfGF, Filename: finetune_training_bi_new_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-qFnnBpNrS3yArIYNB99VCkLr, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-1JS35ObjIzALlwGayDBCGgxr, Filename: finetune_validation_bi_SelfI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-xIw67uv08I8quPgrHZl5F2fg, Filename: finetune_training_bi_new_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-J1oQVVgI6bKUbKVhM1iQdxQz, Filename: finetune_validation_bi_SelfI.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-VcTwfn6KsxtOArnnsk5Bpdul, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-qimrI5DLSyaQQS11b1oBHKIh, Filename: finetune_validation_self_instruct.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-7CKLWSQIOXdJVW0mc78NeT1H, Filename: finetune_train_new_initial.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6gODfDdbxB4ixr8AJHauogam, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-myhSIZsTiI5kGQ33BQmncBUc, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-BaVndSVSR8zvOeZqRXGM85ll, Filename: finetune_validation_selfinstruct_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-SDLdw86GZ01qde2TwGn3S0ur, Filename: finetune_train_augm_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-FBIDLHlcz3gomQ49Snb5TLBX, Filename: finetune_validation_augmented_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-IbBfzQdmJ8EYNSeqXppZ52bn, Filename: finetune_train_self_instruction_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-cRBSeeajdClDoFVHW5rVL3ob, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-TZzPdgvtV5wwwxdUR4bpSyrt, Filename: finetune_validation_augmented_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-YgA2HzdsmzsjikHLjyEpxFwP, Filename: finetune_train_self_instructio_full.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-RIfIxHYcQbBMqn8br0Adqx5x, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-SSwmQm1u8Lz46g7LabY04ri4, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-FdfvPStcKXtC1ZhZd7mefs8R, Filename: finetune_validation_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-2qRDRdUMIupyieWuAoQ3pgpO, Filename: finetune_train_only_self_instruction.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Mg8czMOMDEjWaMQ2MDJw8b4M, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-YyzkVq8j7M33Id6RRUdqlsU0, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-A2penLjZiviSGvqSdX3CKGnS, Filename: finetune_validation_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-Fg0zMmI0TTlaDdsmk6XPzESw, Filename: finetune_train_only_self_instruction.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-7xzZZezgaIbEEcg6n9ipT4WP, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-tw98fjmLD9OybV8hyGy51jLX, Filename: finetune_validation_initial_subset.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-UltciEj04Arj5qFWzMtCZGkB, Filename: finetune_train_only_self_instruction.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-MwpB3saYcvJL2oa4sW2XfQle, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-KXL13xkpEqhMz1oCNcmbLB2c, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-4Fq4ZnfXBFu4dd5GUfROSwCt, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Q5NusqapTVl7yRKuEcTOmLUv, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-e7T6GgaSQryrZ8WlA0WrQ0ZD, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-zzqiq2cjsqNfXo5Q7cBPxuez, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-DZ4nTXuyLFfOHAWU2DtZFLQD, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-lr7cibRK54zNUg3RcvnVtuAN, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-O2s5Gc1vAlMsYLpZhxttY8uq, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-3xdxc0fdeh9N18Xfce8Xohjo, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-uxpAwxMtTKGrERqTSCBA2L9X, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-PrLjnj28e9CA25esHWQMPSmb, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-xapba0m8CiOAK2PpEP200kpj, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-CdBXajY0D2rg1PWavpJlVAVZ, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-3gWyt3eRS3rGHorNZSa5zSkz, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-WZfCIjiHNIKElW28kQ58tdyW, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-TdOTIFPzw29yezhTl9w0bWGM, Filename: finetune_validation_with_self_instruct.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-k8k8busnYErD66CcR0JJjA1j, Filename: finetune_training_with_self_instruct.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-UsHN39kmIS6RxINuLnmriymL, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-GCAbtqTaJiHsbbaO449luoHq, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-QIiCso9yEtkeIbKciCo7yTX8, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-5CqRzMRmqP5pOqCSWe3HolHp, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-lda6U9jCmN4W530aTbJNL2Up, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-u7KjjEQOLEIWFW08VIUZVPsw, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-PW6x7RdMwyOqOsbyLmcfIgI3, Filename: finetune_validation.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-wh6OtrkTRQZKHJEC7W1V1zc9, Filename: finetune_training.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6ma1o8vXyxbvJyQB6s2lDMY2, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-YuW3Iwx67MpEnC7tUXuiMZFN, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-ORABZHgqyCOMn7wds0jzAB05, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-79i8edO5uosZ60rPkKG2SHfT, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-8naTnYs14ijfti9ZHh6cMmJ7, Filename: training_data.json, Purpose: fine-tune, Status: processed\n",
      "ID file-WbsqZhQf4bkK5Udcj6Sk03xV, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-IwxHiCwxTaPB1qPAXQrffR0w, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-RTKaPMnCmlmRa4OWhjLXOTqT, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-UICYrp1IfYrQPwsFtJuJn6qq, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-y2KCLYNGROo5esu3susvUfrr, Filename: training_data.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-49OIolYEKaExdkyFiLnAyZXJ, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-PfRi0tykLlClSKOuRiWxmweM, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-3pgQaz1Xq3eok20xosmPdLay, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-ZE0yhoSN0Q35IjNSA66tX90Q, Filename: self_instruct.json, Purpose: fine-tune, Status: processed\n",
      "ID file-O1SdyKIswZMviNo79kSmzKRk, Filename: self_instruct.json, Purpose: fine-tune, Status: processed\n",
      "ID file-No8fAmof6yjZgvwRCMTmnw3d, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-wosFaNN1KyBjY5r5hTE8Z8Zo, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-t33OqSQXocmpAu4cVGZU1gez, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-m4NqxRfbCphFqCoPi1AEcD4e, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-pYSxeNNOLR09bsLdYDFoxX2w, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-dzpz6H3HjboamdSn2UQbibGZ, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Z7uLfDk6vr1n0ZJ2fAX57A3n, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-GO32waQeLGQSrKnD65RGYm8u, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-mafg4XNg6mkdjFLsdKrKqi0M, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-PhtKvHb4NFcVY8KbcBmccGtV, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-qQdvyHrkcQL5RpxmKe4xWXnm, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-M2ianvGilYyz7YCc3W6WCLEX, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-kn8nJrVYgpq4yBsJdVy1o3tB, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-AmdQuqaYyL3hubageSZFfab4, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-M9Q7ePxoNrRzkdqpLRT7s5iZ, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-x2KkpJnuQfC7l9rd7a5LnHUK, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-jTXCGY1eaqCxSHlEPuzzNiVA, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-aor4HEAlXdRMWryApifiVG9Z, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-LrgMFpLGuiLhz9BlFKZQrDMq, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-9ldOIu9MgxPyGjD2m2ZizWTP, Filename: val_df.json, Purpose: fine-tune, Status: processed\n",
      "ID file-Xvs9uAIuhWEw4foqFDIeLoKb, Filename: self_instruct_and_val_df.json, Purpose: fine-tune, Status: processed\n",
      "ID file-43M77Vvio4fGYwfeabxD6E5J, Filename: self_instruct.json, Purpose: fine-tune, Status: processed\n",
      "ID file-r8fQmGUS1SylNaXcGDXugBl1, Filename: self_instruct_and_val_df.json, Purpose: fine-tune, Status: processed\n",
      "ID file-CNs0xsYWvN9HcGKmwHUX85ov, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-riNhnZ5iQIid4ChoT0tlP8p4, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-xgvkuz2QnHkQOfiVcN9EkRe9, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-udNmyHeuRHhIDv7EPqmCKQ68, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-rkWhD7j2xj50UAk8sJfEQ3Jn, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-6Rm20CbyRs9zyx6dc7NBRibL, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-88OR1EwarZfGeQNyBpU4alDH, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-udp7Mog0cM9x0yjSQ8QCmnMg, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-bE9IwpANDBLGTIFrkj4T1v0R, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-NiqCcybnAohmkHDKRZDzOezB, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-qoKOFjUpNlZ6mbhbVE5qczAX, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-H8ErMAd6JMV8r43guEpvRLgF, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-OGPw2oP70EOfd39ETtNJJWZE, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Y61lO6RkvDiSYmQxSjYdAxeA, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-7gLe60LVjpjkXv0cwp08GhHv, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-FQtf4aEeig8yla2Dv5jSCdSV, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-9t8Bz8qZQbtFMEAWW4WfjZBP, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-KDAB735l2vdmpNWvOTr67cwd, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-taZ7qjoBidlb0HKSfZMmCLHt, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-3KGzSDeOwEz4BmG8KdJF0m2w, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-W89aXHOw8EQpZFHCAJeydK4N, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-pIdi8SiqT0wgMOuVWVC2EAwy, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-NP8xMW8G1U2HMHJWAVymi9CB, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-vP0fHyImL5NWzOvwUdrXKBE2, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-sRX9KiXYNqSnIxtVAhJFFEuB, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-MYwg9faTXi9NCfxSxA6InuPT, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-xNYBCd7FxFFeNaGq88A5ug19, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-d01ma6nDs7u5Juh0GfEU93I1, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-FpHOlWwCbgeHZEvVZrxZhVe7, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-bC1MKoOtMLAyYEdnFtIYJYlf, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-hkeEE5OT2svVIerbf2rT9XmE, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-Z5O9bGeU3SuUG50NPjOVCMv4, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-8YYl75kpMSrEU3f7zue8Zxut, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-YKaEe3w4ORrxHSgKtuBAacq3, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-46SKfiQZJdnQQfjZuJI3k8ql, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-oxY9aJQCcCeSiWlT1xv84e7Y, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-0AchWO7WLjtUltTOn1uspbrT, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-jTtY7Jj3SN8tyvk2pu4tecH0, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-41oj3RSLmavboYdfjaj8k3gq, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-qzh2QXeiYSFF2EY3av4r2C6H, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-nObPB8NOSKXWNP8tZxs2wovu, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-DgNn6YdHaiJX3LUMNOf9tiRY, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-o0z3oMVhl67Ddoc2btf0q7CA, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-WuAjTOTQjvkmO9rKEvWH9WPZ, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-GTUe4QdiJKXjyGEXpre19fa1, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Hd0IsCwFCLRPZxKIufgVppuf, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-X39R8g0SgC0wwA7UlJRZ6ma1, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-hleZmO1BoIDNiX8M7hLygQXM, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-WnfNyPTj6vCw6GxOECCbnkev, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-zle6sw8fJdo0taeGJmZtR5EK, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-L88jeIY7jOrBguKiZgaSXnRt, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-f8tHVy65A0AMc4s0u4meCJpX, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-rljuwh8aiv9l7zOR0MQhgVyc, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-WbHfazaTJEPOmmwbpM2M2mCd, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-PmFbiLMUrjvNMzP7NqK65Oqm, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-J809BUB5AF9PaQKaaXB6XnYm, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-QRHCSg9zFRLAnFGjLZrtM1Of, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-VEmLUseH5L41anjrAQ7Q4JrY, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-tETHHgzOcgRosgBZ7O35Gy4v, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-4jvyjzXncJZWGp6PObeU8peA, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-cG0hxCCkdSBl0RTFdnKfe6Ih, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Se4RJAp0izhHMlTXwQNlV1Jf, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-YrIcXqh3605UN7yYuiWu1ZSP, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-1TVtl0FDV7BrdFCucDUMriHE, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-LnMaNdskol238FtCx7FquN9s, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-E9GFsfPdUEEiJIHSRbms7QH9, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-QRdA0JUubfYYCSe2fWFzwaNH, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-K1pUPO0vJFIgZ7H6Du1qThzD, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-CB0nDXZ2DiVpzubI3XtUOnUI, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-r3VhZ1HYKs7vw6sWCPdWrx4C, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-9w43FJymSg7f0RAdP7j5iYeP, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-oUUuyCZqAdmgFulwho1NRPsE, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-HfHJcRKG9De2oB4j8YOtXhCe, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-4VArTnnSG2u4JOYXBZ6YqtPu, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-E1eGTtKvzHfb9fxj5a5F3AVg, Filename: chatgpt_annotations_test_ds_1__task_1_train_set_10.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-ijSmS4fR1C6lCEy6NnQaeGi0, Filename: chatgpt_annotations_test_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-DVb9FzP1WhkNYhjbeXUAbhQ6, Filename: chatgpt_annotations_test_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-aMJRdbrcFGAaAcuLtsirbrDk, Filename: chatgpt_annotations_test_ds_1__task_1_eval_set.jsonl, Purpose: fine-tune, Status: processed\n",
      "ID file-6Kels6BjwgbJWxojHvgiaiDf, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-mKm9AU15t4AriP5QXnC2q4P7, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-54jJZWYV1jiq4r7DPNaIvnhC, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-MfE4vwkQQlmw95SpcPvmNyS4, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-y9vGjCPiD4q9vkfv6piZ5GxP, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-AqVtPvWu1Ko3Eet6TYvvXxNh, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-kg4L1ROygRLzB3Lest2xCDhy, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-smFrheeSyJKibAjCEylUPe4o, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-Ru91h1BQl29iB6m2PXiXvHwX, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-dQpT4V8lPCtEXbyq4bMhPB9e, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-MZc6vLUKkqH3jck89tZsy2Mv, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-mpsEW0NC7jqp1kXOu3pRo9BR, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-E1vpI90e56d2AxtDVXjO42fH, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-pjthj0QzeRIieqeaqYDX2iwg, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-JjIQEZHt6B7zXC9RWdbm2NmR, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-i76SiJTNqW90ryemLheMVhOh, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-MmAALoV7Nni1RqwteOswPZY8, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-5o5kM1QqHrhKzGXHCNooz1ok, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-8FICQR9b8SmcXR3ZyPwZOxcx, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-RANuO9DcxzIxadpvX9exJeth, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-33FxH03JzQoF0xkqNfsc3Zuc, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-wB20N8yumLPGupiRvxR7YYfd, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-8gJnWYLX1e9Bxo0BJWBAF51s, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-CrIEHnu9MDyPxPS968JOMMOv, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-66UPQsCcIrUXX09NsN7dKr9l, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-ucgxNxFbLSzMaF9vB29WMxAb, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-ISi2yhmpl4CEUt00lHdAtq4c, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-n9peaaHuHrbOcwfQLXqkMueb, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-9atMTPwtrNc1N8HIqcSO13oI, Filename: step_metrics.csv, Purpose: fine-tune-results, Status: processed\n",
      "ID file-n4hROkpk30nWQDwtFGBqp2UZ, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-atMANf0QT3M8PnWodCFDlSFx, Filename: file, Purpose: fine-tune, Status: processed\n",
      "ID file-riaTSACDmRO5NuqG9rf65lNP, Filename: file, Purpose: fine-tune, Status: processed\n"
     ]
    }
   ],
   "source": [
    "files = client.files.list()\n",
    "\n",
    "for file in files.data:\n",
    "    print(f\"ID {file.id}, Filename: {file.filename}, Purpose: {file.purpose}, Status: {file.status}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "6b0baf56",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Job ID: ftjob-7tKvuoKjuGjGnWN3xHvW2QJt\n",
      "Status: validating_files\n"
     ]
    }
   ],
   "source": [
    "# Section 19\n",
    "model_name = (df_name.replace('__', '_')\n",
    "              .replace('train_set', 'trn')\n",
    "              .replace('task', 't')\n",
    "              .replace('_single_letter_labels', '_sl'))\n",
    "job_id = fine_tune_chat_gpt(evaluation_file_id, training_file_id, model_name=model_name, n_epochs=n_epochs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "6aff8caf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The model ft:gpt-4o-mini-2024-07-18:university-of-zurich:ds-4-t-1-trn-250:B5rboCOG has been successfully fine-tuned\n"
     ]
    }
   ],
   "source": [
    "# Section 20\n",
    "# Print the model name\n",
    "response = client.fine_tuning.jobs.retrieve(job_id)\n",
    "response_dict = response.to_dict()\n",
    "full_model_name = response_dict[\"fine_tuned_model\"]\n",
    "print(f'The model {full_model_name} has been successfully fine-tuned')\n",
    "wandb.config['model_name_openai'] = full_model_name\n",
    "wandb.config['finetuning_jobid'] = job_id\n",
    "wandb.config['training_file_openai_id'] = response_dict['training_file']\n",
    "wandb.config['validation_file_openai_id'] = response_dict['validation_file']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "45b169bf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Step 48/63: training loss=0.05, validation loss=0.17\n",
      "Step 49/63: training loss=0.02, validation loss=0.03\n",
      "Step 50/63: training loss=0.03, validation loss=0.07\n",
      "Step 51/63: training loss=0.03, validation loss=0.11\n",
      "Step 52/63: training loss=0.03, validation loss=0.08, full validation loss=0.09\n",
      "Step 53/63: training loss=0.09, validation loss=0.14\n",
      "Step 54/63: training loss=0.03, validation loss=0.05\n",
      "Step 55/63: training loss=0.04, validation loss=0.08\n",
      "Step 56/63: training loss=0.04, validation loss=0.15\n",
      "Step 57/63: training loss=0.06, validation loss=0.02\n",
      "Step 58/63: training loss=0.01, validation loss=0.10\n",
      "Step 59/63: training loss=0.02, validation loss=0.04\n",
      "Step 60/63: training loss=0.03, validation loss=0.11\n",
      "Step 61/63: training loss=0.01, validation loss=0.06\n",
      "Step 62/63: training loss=0.02, validation loss=0.08\n",
      "Step 63/63: training loss=0.02, validation loss=0.01, full validation loss=0.07\n",
      "Checkpoint created at step 39\n",
      "Checkpoint created at step 52\n",
      "New fine-tuned model created\n",
      "The job has successfully completed\n"
     ]
    }
   ],
   "source": [
    "# Section 21\n",
    "# Print the events (training history of the model)\n",
    "print_and_log_finetuning_event_history(job_id)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2c579b37",
   "metadata": {},
   "source": [
    "## Evaluate Model on Validation Set with Annotations of the Group from which the Training Data is!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "fba537b1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "##################################################\n",
      "Getting predictions on the evaluation set\n"
     ]
    }
   ],
   "source": [
    "# Section 22\n",
    "# Evaluate the model on the evaluation set and store the predictions\n",
    "print(\"\\n\" + \"#\" * 50)\n",
    "print(\"Getting predictions on the evaluation set\")\n",
    "predictions = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "0a544a5b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 394/394 [04:02<00:00,  1.63it/s]\n"
     ]
    }
   ],
   "source": [
    "# Section 23\n",
    "for messages in tqdm(eval_df['openai_instance_without_completion'].tolist()):\n",
    "    # Retry the completion at least COMPLETION_RETRIES times\n",
    "    num_retries = 2\n",
    "    response = None\n",
    "    while num_retries < COMPLETION_RETRIES and response is None:\n",
    "        try:\n",
    "            response = client.chat.completions.create(\n",
    "                model=full_model_name,\n",
    "                messages=messages,\n",
    "                temperature=temp,\n",
    "                n=1\n",
    "            )\n",
    "        except Exception as e:\n",
    "            print('Error getting predictions. Retrying...')\n",
    "            time.sleep(5)\n",
    "            num_retries += 1\n",
    "            if num_retries >= COMPLETION_RETRIES:\n",
    "                print('Maximum amount of retires reached')\n",
    "                raise e\n",
    "    response_dict = response.to_dict()\n",
    "    predictions.append(response_dict['choices'][0]['message']['content'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "337b13ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 24\n",
    "# Add predictions to df\n",
    "eval_df['prediction'] = predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "fe242636",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 25\n",
    "# Store output\n",
    "predictions_output_dir = os.path.join(output_dir, 'predictions',\n",
    "                                      f'dataset_{dataset}_task_{task}')\n",
    "os.makedirs(predictions_output_dir, exist_ok=True)\n",
    "datasets[eval_set_name].to_csv(\n",
    "    os.path.join(predictions_output_dir, f\"{model_name}-{run_name}.csv\"),\n",
    "    index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "bcaddde5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 26\n",
    "# Get performance metrics\n",
    "y_true = eval_df['completion_label']\n",
    "y_pred = eval_df['prediction']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "3d085eda",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 27\n",
    "label_type = 'full_name' if not not_use_full_labels else 'short_name'\n",
    "display_labels = task_to_display_labels[task][label_type]\n",
    "labels = display_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "a1ce1492",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'HATE SPEECH': {'f1-score': 0.8065573770491803,\n",
      "                 'precision': 0.7365269461077845,\n",
      "                 'recall': 0.8913043478260869,\n",
      "                 'support': 138},\n",
      " 'KEINE HATE SPEECH': {'f1-score': 0.8853932584269664,\n",
      "                       'precision': 0.9292452830188679,\n",
      "                       'recall': 0.8454935622317596,\n",
      "                       'support': 233},\n",
      " 'TOXIC SPEECH': {'f1-score': 0.5789473684210527,\n",
      "                  'precision': 0.7333333333333333,\n",
      "                  'recall': 0.4782608695652174,\n",
      "                  'support': 23},\n",
      " 'accuracy': 0.8401015228426396,\n",
      " 'macro avg': {'f1-score': 0.7569660012990664,\n",
      "               'precision': 0.7997018541533286,\n",
      "               'recall': 0.7383529265410212,\n",
      "               'support': 394},\n",
      " 'weighted avg': {'f1-score': 0.8398917175633357,\n",
      "                  'precision': 0.8503084674440029,\n",
      "                  'recall': 0.8401015228426396,\n",
      "                  'support': 394}}\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1500x500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Section 28\n",
    "cm_plot, classification_report, metrics = plot_count_and_normalized_confusion_matrix(\n",
    "    y_true, y_pred, display_labels, labels, xticks_rotation='horizontal')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "75c64dc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 29\n",
    "# Log metrics\n",
    "for metric_name, metric_value in metrics.items():\n",
    "    wandb.log({metric_name: metric_value})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "4aaa4209",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Section 31\n",
    "# Log the confusion matrix matplotlib figure\n",
    "wandb.log({'confusion_matrix': wandb.Image(cm_plot)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "45aaca05",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Artifact classification_report_ds_4_t_1_trn_250>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Section 32\n",
    "# Log the classification report as an artifact\n",
    "classification_report = (pd.DataFrame({k: v for k, v in classification_report.items() if k != 'accuracy'})\n",
    "                         .transpose().reset_index())\n",
    "\n",
    "wandb.log({'classification_report': wandb.Table(\n",
    "    dataframe=classification_report)})\n",
    "\n",
    "classification_report_artifact = wandb.Artifact(\n",
    "      f'classification_report_{model_name}', type='classification_report')\n",
    "\n",
    "with classification_report_artifact.new_file('classification_report.txt', mode='w') as f:\n",
    "    f.write(pprint.pformat(classification_report))\n",
    "\n",
    "wandb.run.log_artifact(classification_report_artifact)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6db7bbf2",
   "metadata": {},
   "source": [
    "## Evaluation on Second Evaluation Set\n",
    "In this section, we will evaluate the performance of our model on a second evaluation set (`second_eval_set`)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "8cbb7ff7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['ds_4__task_1_eval_set', 'ds_x__task_1_full_eval', 'ds_4__task_1_train_set_250'])\n"
     ]
    }
   ],
   "source": [
    "print(datasets.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "5dd00a6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "second_eval_set_name = f'ds_{dataset_eval}__task_{task}_full_eval'\n",
    "full_eval_df = datasets[second_eval_set_name]\n",
    "for df_name, df in datasets.items():\n",
    "    if df_name == second_eval_set_name:\n",
    "        continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "9cfc903c",
   "metadata": {},
   "outputs": [],
   "source": [
    "full_eval_df = datasets[second_eval_set_name]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "303aeed5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "##################################################\n",
      "Getting predictions on the evaluation set\n"
     ]
    }
   ],
   "source": [
    "# Load the second evaluation set (second_eval_set)\n",
    "# Evaluate the model on the evaluation set and store the predictions\n",
    "print(\"\\n\" + \"#\" * 50)\n",
    "print(\"Getting predictions on the evaluation set\")\n",
    "predictions_2 = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "7c218926",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 494/494 [04:11<00:00,  1.97it/s]\n"
     ]
    }
   ],
   "source": [
    "for messages in tqdm(full_eval_df['openai_instance_without_completion'].tolist()):\n",
    "    # Retry the completion at least COMPLETION_RETRIES times\n",
    "    num_retries = 2\n",
    "    response = None\n",
    "    while num_retries < COMPLETION_RETRIES and response is None:\n",
    "        try:\n",
    "            response = client.chat.completions.create(\n",
    "                model=full_model_name,\n",
    "                messages=messages,\n",
    "                temperature=temp,\n",
    "                n=1\n",
    "            )\n",
    "        except Exception as e:\n",
    "            print('Error getting predictions. Retrying...')\n",
    "            time.sleep(5)\n",
    "            num_retries += 1\n",
    "            if num_retries >= COMPLETION_RETRIES:\n",
    "                print('Maximum amount of retires reached')\n",
    "                raise e\n",
    "    response_dict = response.to_dict()\n",
    "    predictions_2.append(response_dict['choices'][0]['message']['content'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "789fce14",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add predictions to df\n",
    "full_eval_df['prediction'] = predictions_2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "616c4658",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions_output_dir_2 = os.path.join(output_dir, 'predictions_2',\n",
    "                                      f'dataset_{dataset}_task_{task}')\n",
    "os.makedirs(predictions_output_dir_2, exist_ok=True)\n",
    "datasets[second_eval_set_name].to_csv(\n",
    "    os.path.join(predictions_output_dir_2, f\"{model_name}-{run_name}.csv\"),\n",
    "    index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "ec8481ac",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get performance metrics\n",
    "y_true_2 = full_eval_df['completion_label']\n",
    "y_pred_2 = full_eval_df['prediction']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "296deb14",
   "metadata": {},
   "outputs": [],
   "source": [
    "label_type_2 = 'full_name' if not not_use_full_labels else 'short_name'\n",
    "display_labels_2 = task_to_display_labels[task][label_type]\n",
    "labels_2 = display_labels_2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "83dbd847",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'HATE SPEECH': {'f1-score': 0.64,\n",
      "                 'precision': 0.5194805194805194,\n",
      "                 'recall': 0.8333333333333334,\n",
      "                 'support': 144},\n",
      " 'KEINE HATE SPEECH': {'f1-score': 0.7720000000000001,\n",
      "                       'precision': 0.7909836065573771,\n",
      "                       'recall': 0.75390625,\n",
      "                       'support': 256},\n",
      " 'TOXIC SPEECH': {'f1-score': 0.24778761061946902,\n",
      "                  'precision': 0.7368421052631579,\n",
      "                  'recall': 0.14893617021276595,\n",
      "                  'support': 94},\n",
      " 'accuracy': 0.6619433198380567,\n",
      " 'macro avg': {'f1-score': 0.5532625368731564,\n",
      "               'precision': 0.6824354104336848,\n",
      "               'recall': 0.5787252511820332,\n",
      "               'support': 494},\n",
      " 'weighted avg': {'f1-score': 0.633773351008563,\n",
      "                  'precision': 0.7015387772846562,\n",
      "                  'recall': 0.6619433198380567,\n",
      "                  'support': 494}}\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABQYAAAHUCAYAAABoGvWaAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAA9hAAAPYQGoP6dpAACzI0lEQVR4nOzdd3RU1drH8d+kJ6SRBFIgkNCRXqSqgFKMgA2RIlIEFKMiAoKIAiIXFBGwgaJUlaKCDRBEqV6QS5eidEiAhBJKIJB+3j94MzhMAhmYMCnfz1pnLWafffbZZ5KQJ8/sYjIMwxAAAAAAAACAIsXJ0R0AAAAAAAAAcOeRGAQAAAAAAACKIBKDAAAAAAAAQBFEYhAAAAAAAAAogkgMAgAAAAAAAEUQiUEAAAAAAACgCCIxCAAAAAAAABRBJAYBAAAAAACAIojEIAAAAAAAAFAEkRgEYHd//fWXevXqpcjISHl4eMjb21t169bV+PHjdfbsWUd3T5I0d+5cTZ482aZrMjMz9eWXX6ply5YKCgqSq6urSpYsqXbt2unnn39WZmZm3nT2/3300UeqUKGC3NzcZDKZdP78ebu2P2vWLJlMJh05csSu7eZG8+bNZTKZVK5cORmGYXV+7dq1MplMMplMmjVrls3tnzhxQqNGjdL27dttuq5nz56KiIiw+X4AADhC1u9yDw8PHT161Op88+bNVb16dQf0zD6y+70cERGhnj173tF+HDlyxKaY5NChQ3rxxRdVqVIleXp6ysvLS9WqVdMbb7yh48eP53lf27Ztq4CAAJlMJg0YMMDu93DE10CSVq9efdP48P7775fJZLrleO5W/maw9fsDcDQXR3cAQOHy+eefKzo6WpUrV9arr76qu+66S2lpadq8ebM+/fRTbdiwQd9//72ju6m5c+dq165duQ6OkpOT9eijj+rXX39V586dNXXqVIWEhOj06dNatmyZOnbsqAULFuiRRx7Jk/5u375d/fv3V58+fdSjRw+5uLjIx8fHrvdo27atNmzYoNDQULu2m1s+Pj46fPiwVq5cqQceeMDi3IwZM+Tr66vExMRbavvEiRN66623FBERodq1a+f6ujfffFMvv/zyLd0TAABHSUlJ0RtvvKEvv/zS0V3Jc99//718fX0d3Y0cLV68WJ07d1ZQUJBefPFF1alTRyaTSTt37tSMGTO0ZMkSbdu2Lc/u/8orr2jjxo2aMWOGQkJC8iTOc/TXwMfHR9OnT7dKTh4+fFirV6++rb7Z+jeDJIWGhmrDhg0qX778Ld8XuJNIDAKwmw0bNuj5559Xq1at9MMPP8jd3d18rlWrVho0aJCWLVvmwB7euoEDB2r58uWaPXu2unfvbnHu8ccf16uvvqorV67k2f13794tSerbt68aNGiQJ/coUaKESpQokSdt50aZMmXk4+OjGTNmWCQGL168qG+//VZPPfWUPv/88zvSl8uXL8vLy4uADgBQID344IOaO3euBg8erFq1auXZfa5cuSJPT888az836tSp49D738jhw4fVuXNnVapUSatWrZKfn5/53P3336/+/fvn+Qfmu3btUoMGDfToo4/m2T0c/TXo1KmTvvjiC+3fv18VK1Y0l8+YMUOlSpVSjRo1tGfPnjzvR0ZGhtLT0+Xu7q5GjRrl+f0Ae2EqMQC7GTt2rEwmk6ZNm2aRFMzi5uamhx9+2Pw6MzNT48ePV5UqVeTu7q6SJUuqe/fuOnbsmMV1OU1PaN68uZo3b25+nTWdYN68eRo+fLjCwsLk6+urli1bau/evRbXLVmyREePHjVPPzCZTDk+V3x8vL744gu1adPGKimYpWLFiqpZs6b5dUxMjLp166aSJUvK3d1dVatW1fvvv28x3ThrmsGECRM0ceJERUZGytvbW40bN9aff/5p0d9u3bpJkho2bCiTyWR+P3L73mRmZmrMmDGqXLmyPD095e/vr5o1a+qDDz4w18lpKvGMGTNUq1YteXh4KCAgQI899pj+/vtvizo9e/aUt7e3Dhw4oIceekje3t4KDw/XoEGDlJKSkuN7e71nnnlGixYtspgmPX/+fElS586dreofOHBAvXr1UsWKFeXl5aVSpUqpffv22rlzp7nO6tWrdffdd0uSevXqZf56jxo1yqLvO3fuVOvWreXj42NOTF4/ZWn+/PkymUz6+OOPLfoxcuRIOTs7a8WKFbl+VgAA8sqQIUMUGBiooUOH3rRucnKyhg0bpsjISLm5ualUqVJ64YUXrJYsiYiIULt27bRo0SLVqVNHHh4eeuutt8zx19y5czV06FCFhobK29tb7du318mTJ3Xx4kU9++yzCgoKUlBQkHr16qVLly5ZtP3JJ5/ovvvuU8mSJVWsWDHVqFFD48ePV1pa2k37f30slLU8SXbHv6d2xsfH67nnnlPp0qXl5uamyMhIvfXWW0pPT7do/8SJE3ryySfl4+MjPz8/derUSfHx8TftlyRNnDhRSUlJmjJlikVSMIvJZNLjjz9uUWavuCvr63LgwAH98ssv5vfgyJEjOcZ8WdesXr3aXLZt2za1a9fOHNOGhYWpbdu2FvF6dvGoPWPhm2nVqpXCw8M1Y8YMc1lmZqZmz56tHj16yMnJOu2Rm++5G/3NkNX38ePHa8yYMYqMjJS7u7tWrVplNZU4OTlZderUUYUKFXThwgVz+/Hx8QoJCVHz5s2VkZGR6+cF7I0RgwDsIiMjQytXrlS9evUUHh6eq2uef/55TZs2TS+++KLatWunI0eO6M0339Tq1au1detWBQUF3VJfXn/9dTVt2lRffPGFEhMTNXToULVv315///23nJ2dNWXKFD377LM6ePBgrj6lXbVqldLS0nL9Sevp06fVpEkTpaam6u2331ZERIQWL16swYMH6+DBg5oyZYpF/U8++URVqlQxr1/y5ptv6qGHHtLhw4fl5+enKVOmaN68eRozZoxmzpypKlWq2Dyyb/z48Ro1apTeeOMN3XfffUpLS9M///xz03UKx40bp9dff11dunTRuHHjlJCQoFGjRqlx48batGmTxaeyaWlpevjhh9W7d28NGjRIa9eu1dtvvy0/Pz+NGDEiV/3s3LmzXnnlFc2bN0/PP/+8JGn69Ol64oknsp0GcuLECQUGBuqdd95RiRIldPbsWc2ePVsNGzbUtm3bVLlyZdWtW1czZ85Ur1699MYbb6ht27aSpNKlS5vbSU1N1cMPP6znnntOr732mtUfBf/u35o1azRo0CA1atRI9evX18qVKzVmzBi9/vrratWqVa6eEwCAvOTj46M33nhDL7/8slauXKn7778/23qGYejRRx/V77//rmHDhunee+/VX3/9pZEjR2rDhg3asGGDxYe9W7du1d9//6033nhDkZGRKlasmJKSkiRdjb9atGihWbNm6ciRIxo8eLC6dOkiFxcX1apVS/PmzdO2bdv0+uuvy8fHRx9++KG53YMHD6pr167m5OSOHTv0n//8R//8849Fsic3pkyZYrX0yJtvvqlVq1apcuXKkq4mZBo0aCAnJyeNGDFC5cuX14YNGzRmzBgdOXJEM2fOlHR1RGTLli114sQJjRs3TpUqVdKSJUvUqVOnXPXl119/VXBwcK5Hj9kz7qpbt642bNigxx57TOXLl9eECRMkyaapxElJSWrVqpUiIyP1ySefKDg4WPHx8Vq1apUuXryY43X2joVvxsnJST179tT06dM1ZswYOTs769dff9WxY8fUq1evbJeFyc33XG7+Zvjwww9VqVIlTZgwQb6+vhZfoyweHh765ptvVK9ePT3zzDNauHChMjMz9dRTT8kwDM2bN0/Ozs43fU4gzxgAYAfx8fGGJKNz5865qv/3338bkozo6GiL8o0bNxqSjNdff91cVrZsWaNHjx5WbTRr1sxo1qyZ+fWqVasMScZDDz1kUe+bb74xJBkbNmwwl7Vt29YoW7Zsrvr6zjvvGJKMZcuW5ar+a6+9ZkgyNm7caFH+/PPPGyaTydi7d69hGIZx+PBhQ5JRo0YNIz093Vzvf//7nyHJmDdvnrls5syZhiRj06ZNFm3m9r1p166dUbt27Rv2O+sehw8fNgzDMM6dO2d4enpavZ8xMTGGu7u70bVrV3NZjx49DEnGN998Y1H3oYceMipXrnzD+2b1t1q1aua26tevbxiGYezevduQZKxevdrYtGmTIcmYOXNmju2kp6cbqampRsWKFY1XXnnFXH6ja7P6PmPGjGzPXf99kpycbNSpU8eIjIw09uzZYwQHBxvNmjWz+BoCAOAI/44XUlJSjHLlyhn169c3MjMzDcOw/H1rGIaxbNkyQ5Ixfvx4i3YWLFhgSDKmTZtmLitbtqzh7OxsjmOyZMVf7du3tygfMGCAIcno37+/Rfmjjz5qBAQE5PgMGRkZRlpamjFnzhzD2dnZOHv2rPlcdr+Xc4qFsrz33ntWz/Lcc88Z3t7extGjRy3qTpgwwZBk7N692zAMw5g6daohyfjxxx8t6vXt2/emMYlhGIaHh4fRqFGjG9bJkldxV9myZY22bdtalF0f82XJ+lquWrXKMAzD2Lx5syHJ+OGHH27Y9+u/BnkRC2cnq7/ffvutcejQIcNkMhmLFy82DMMwOnbsaDRv3twwjJvH/Tf6nsvp2qy+ly9f3khNTc323PXfH1k/V5MnTzZGjBhhODk5Gb/++usNnxG4E5hKDMAhVq1aJUlW0w4aNGigqlWr6vfff7/ltv89XVmSeYpvdrvz5YWVK1fqrrvusloLsGfPnjIMQytXrrQob9u2rcWnhHnR3wYNGmjHjh2Kjo7W8uXLc7WJx4YNG3TlyhWrr1F4eLjuv/9+q6+RyWRS+/btLcpq1qxp83M888wz2rx5s3bu3Knp06erfPnyuu+++7Ktm56errFjx+quu+6Sm5ubXFxc5Obmpv3791tNu7mZDh065Kqeu7u7vvnmGyUkJKhu3bp80gsAyJfc3Nw0ZswYbd68Wd988022dbJikut/13fs2FHFihWz+l1fs2ZNVapUKdu22rVrZ/G6atWqkmQeqf/v8rNnz1pMJ962bZsefvhhBQYGytnZWa6ururevbsyMjK0b9++mz9sDubNm6chQ4bojTfeUN++fc3lixcvVosWLRQWFqb09HTzERUVJUlas2aNpKvxqo+Pj1Vs2bVr11vuU04cFXfdSIUKFVS8eHENHTpUn376aa7X6XNELBwZGanmzZtrxowZSkhI0I8//qhnnnkmx/r2+p57+OGH5erqmqu6Tz75pJ5//nm9+uqrzDZBvkJiEIBdBAUFycvLS4cPH85V/YSEBEnZT2cICwszn78VgYGBFq+zpsDc6uYgZcqUkSSbni2n58o6/2/27m92hg0bpgkTJujPP/9UVFSUAgMD9cADD2jz5s05XmPr18jLy0seHh4WZe7u7kpOTrapr/fdd58qVqyozz77TF9++aWeeeaZHNeAHDhwoN588009+uij+vnnn7Vx40Zt2rRJtWrVsun98/LysmnHugoVKujee+9VcnKynnrqKYft5AwAwI107txZdevW1fDhw7Ndry8hIUEuLi5WS5SYTCaFhIRY/a6/0e+7gIAAi9dubm43LM+KD2JiYnTvvffq+PHj+uCDD7Ru3Tpt2rRJn3zyiaRbj4dWrVqlnj17qnv37nr77bctzp08eVI///yzXF1dLY5q1apJks6cOSPp6vsTHBxs1XZISEiu+lCmTJk8i43tFXfdiJ+fn9asWaPatWvr9ddfV7Vq1RQWFqaRI0fecP1HR8XCvXv31s8//6yJEyfK09NTTzzxRLb17Pk9Z2sM+MwzzygtLU0uLi7q37+/TdcCeYXEIAC7cHZ21gMPPKAtW7ZYbR6SnawAIC4uzurciRMnLNYX9PDwyHYDi6ygLa+1aNFCrq6u+uGHH3JVPzAwMMfnknTLaydmJ7fvjYuLiwYOHKitW7fq7NmzmjdvnmJjY9WmTRtdvnw527Zt+RrZW69evTR16lSdPXtWPXr0yLHeV199pe7du2vs2LFq06aNGjRooPr169v8vXGjzWey88UXX2jJkiVq0KCBPv74Y23cuNGm6wEAuBNMJpPeffddHTx4UNOmTbM6HxgYqPT0dJ0+fdqi3DAMxcfHW/2ut/X3ZW788MMPSkpK0qJFi9StWzfdc889ql+/vjmBeCv++usvPfroo2rWrJk+//xzq/NBQUFq3bq1Nm3alO3Ru3dvSVffn5MnT1pdn9vNR9q0aaOTJ0/maiONOxl3ZSUUr48hs4ufatSoofnz5yshIUHbt29Xp06dNHr0aL3//vs5tn8nY+F/e/zxx+Xl5aV33nlHnTt3znHHbHt+z9nyM5GUlKSnn35alSpVkqenp/r06WPz/YC8QGIQgN0MGzZMhmGob9++Sk1NtTqflpamn3/+WZLMi2B/9dVXFnU2bdqkv//+27wrrHR1p7O//vrLot6+ffssdhq2lbu7e64/DQwJCVGfPn20fPlyzZkzJ9s6Bw8eNPfxgQce0J49e7R161aLOnPmzJHJZFKLFi1uud/Xu5X3xt/fX0888YReeOEFnT171mpHuiyNGzeWp6en1dfo2LFjWrlypcXXyN569Oih9u3b69VXX1WpUqVyrGcymax2wF6yZImOHz9uUWbPUZg7d+5U//791b17d61bt041a9ZUp06ddO7cudtuGwAAe2vZsqVatWql0aNHW+0GnPW7/Prf9QsXLlRSUlKe/q7PkpVY+ffvc8Mwsk3o5UZMTIyioqJUrlw5LVy4MNtpnu3atdOuXbtUvnx51a9f3+rIGtnWokULXbx4UT/99JPF9XPnzs1VX1555RUVK1ZM0dHRFrvRZjEMw7ypxZ2MuyIiIiTJKoa8/jn/zWQyqVatWpo0aZL8/f2t4tx/u5Ox8L95enpqxIgRat++vXkTu+zY8j1ny98MN9OvXz/FxMRo0aJFmj59un766SdNmjTJLm0Dt4NdiQHYTePGjTV16lRFR0erXr16ev7551WtWjWlpaVp27ZtmjZtmqpXr6727durcuXKevbZZ/XRRx/JyclJUVFR5l2Jw8PD9corr5jbffrpp9WtWzdFR0erQ4cOOnr0qMaPH2/zzrz/VqNGDS1atEhTp05VvXr15OTkpPr16+dYf+LEiTp06JB69uyp5cuX67HHHlNwcLDOnDmjFStWaObMmZo/f75q1qypV155RXPmzFHbtm01evRolS1bVkuWLNGUKVP0/PPP57g2z63I7XvTvn17Va9eXfXr11eJEiV09OhRTZ48WWXLls129zTpagLxzTff1Ouvv67u3burS5cuSkhI0FtvvSUPDw+NHDnSbs9xvbCwsFyN0GzXrp1mzZqlKlWqqGbNmtqyZYvee+89ix2HJal8+fLy9PTU119/rapVq8rb21thYWHmwD+3kpKS9OSTTyoyMlJTpkyRm5ubvvnmG9WtW1e9evXK9ahSAADupHfffVf16tXTqVOnzNNlJalVq1Zq06aNhg4dqsTERDVt2tS8K3GdOnX09NNP53nfWrVqJTc3N3Xp0kVDhgxRcnKypk6dessfuEVFRen8+fP6+OOPtXv3botz5cuXV4kSJTR69GitWLFCTZo0Uf/+/VW5cmUlJyfryJEjWrp0qT799FOVLl1a3bt316RJk9S9e3f95z//UcWKFbV06VItX748V32JjIzU/Pnz1alTJ9WuXVsvvvii6tSpI0nas2ePZsyYIcMw9Nhjj93RuOvuu+9W5cqVNXjwYKWnp6t48eL6/vvv9ccff1jUW7x4saZMmaJHH31U5cqVk2EYWrRokc6fP3/DtfHuZCx8vYEDB2rgwIE3rGPL95ytfzPk5IsvvtBXX32lmTNnqlq1aqpWrZpefPFFDR06VE2bNrVajxG4oxy16wmAwmv79u1Gjx49jDJlyhhubm5GsWLFjDp16hgjRowwTp06Za6XkZFhvPvuu0alSpUMV1dXIygoyOjWrZsRGxtr0V5mZqYxfvx4o1y5coaHh4dRv359Y+XKlTnuSvztt99aXJ/dzmBnz541nnjiCcPf398wmUxGbv47TE9PN2bPnm3cf//9RkBAgOHi4mKUKFHCiIqKMubOnWtkZGSY6x49etTo2rWrERgYaLi6uhqVK1c23nvvPYs6Wf167733rO4lyRg5cqT5dU67Euf2vXn//feNJk2aGEFBQYabm5tRpkwZo3fv3saRI0es7nH9DnVffPGFUbNmTcPNzc3w8/MzHnnkEfNufVl69OhhFCtWzOo5Ro4cmav39vpdErOT3c7C586dM3r37m2ULFnS8PLyMu655x5j3bp1Vs9vGIYxb948o0qVKoarq6vF+5tT37PO/Xsnum7duhleXl5Wz//tt98akoxJkybd9FkBAMgrOcULhmEYXbt2NSRZ/b69cuWKMXToUKNs2bKGq6urERoaajz//PPGuXPnLOplt7utYeQcf+XUl6zY4PTp0+ayn3/+2ahVq5bh4eFhlCpVynj11VeNX375xWKHXMPI3a7EknI8/h1DnD592ujfv78RGRlpuLq6GgEBAUa9evWM4cOHG5cuXTLXO3bsmNGhQwfD29vb8PHxMTp06GCsX78+V7sSZzl48KARHR1tVKhQwXB3dzc8PT2Nu+66yxg4cGCex105fd327dtntG7d2vD19TVKlChhvPTSS8aSJUss3vN//vnH6NKli1G+fHnD09PT8PPzMxo0aGDMmjXL6h7X7wxt71g4Ozl9710vu52Fc/s9l9PfDDfq+/V/e/z111+Gp6en1XuUnJxs1KtXz4iIiLD6eQPuJJNhGEZeJx8BAAAAAAAA5C+sMQgAAAAAAAAUQSQGAQAAAAAAgCKIxCAAAAAAAABQBJEYBAAAAAAAAIogEoMAAAAAAABAEURiEAAAAAAAACiCXBzdAcBeMjMzdeLECfn4+MhkMjm6OwCAO8gwDF28eFFhYWFycsrbzz2Tk5OVmppqt/bc3Nzk4eFht/aAwog4DwCKrjsZ50n2jfUKQpxHYhCFxokTJxQeHu7obgAAHCg2NlalS5fOs/aTk5MVWdZb8acy7NZmSEiIDh8+nO+DRsCRiPMAAHkd50n2j/UKQpxHYhCFho+PjySp3LRX5OTp7uDewBEiRiQ5ugtwsMy4k47uAhwk3UjT2pTvzb8L8kpqaqriT2Xo8Jay8vW5/U+sEy9mKrLeUaWmpubrgBFwtKyf7aNbI+TrzWpIRdHjXTo5ugtwIKf9sY7uAhwo3UjVmovf5HmcJ9k31isocR6JQRQaWdNKnDzd5eyVf3/okHdcnNId3QU4WKbJzdFdgIPdqSmGvj5OdkkMAsidrJ9tX29+9ooqF2fi+6LMiRgPunNxnlS0Yj0SgwAAADbKMDKVYdinHQAAAOQv9oj1CkqcR2IQAADARpkylKnbzwzaow0AAADYlz1ivYIS5xWNcZEAAAAAAAAALDBiEAAAwEaZypQ9JofYpxUAAADYkz1ivYIS55EYBAAAsFGGYSjDuP3pIfZoAwAAAPZlj1ivoMR5TCUGAAAAAAAAiiBGDAIAANiIzUcAAAAKr6K0+QiJQQAAABtlylAGiUEAAIBCyR6xXkGJ85hKDAAAAAAAADjYlClTFBkZKQ8PD9WrV0/r1q27Yf2vv/5atWrVkpeXl0JDQ9WrVy8lJCTYdE8SgwAAADbKml5ijwMAAAD5iyPivAULFmjAgAEaPny4tm3bpnvvvVdRUVGKiYnJtv4ff/yh7t27q3fv3tq9e7e+/fZbbdq0SX369LHpviQGAQAAbJS1U509DgAAAOQvjojzJk6cqN69e6tPnz6qWrWqJk+erPDwcE2dOjXb+n/++aciIiLUv39/RUZG6p577tFzzz2nzZs323RfEoMAAAAAAABAHkhMTLQ4UlJSrOqkpqZqy5Ytat26tUV569attX79+mzbbdKkiY4dO6alS5fKMAydPHlS3333ndq2bWtT/0gMAgAA2CjTjgcAAADyF3vGeeHh4fLz8zMf48aNs7rfmTNnlJGRoeDgYIvy4OBgxcfHZ9vHJk2a6Ouvv1anTp3k5uamkJAQ+fv766OPPrLpWdmVGAAAwEYZdtqV2B5tAAAAwL7sEetlXR8bGytfX19zubu7e47XmEwmi9eGYViVZdmzZ4/69++vESNGqE2bNoqLi9Orr76qfv36afr06bnuJ4lBAAAAAAAAIA/4+vpaJAazExQUJGdnZ6vRgadOnbIaRZhl3Lhxatq0qV599VVJUs2aNVWsWDHde++9GjNmjEJDQ3PVP6YSAwAA2CjDsN8BAACA/OVOx3lubm6qV6+eVqxYYVG+YsUKNWnSJNtrLl++LCcny7Ses7OzpKsjDXOLEYMAAAA2stf6gKwxCAAAkP/YI9az9fqBAwfq6aefVv369dW4cWNNmzZNMTEx6tevnyRp2LBhOn78uObMmSNJat++vfr27aupU6eapxIPGDBADRo0UFhYWK7vS2IQAAAAAAAAcKBOnTopISFBo0ePVlxcnKpXr66lS5eqbNmykqS4uDjFxMSY6/fs2VMXL17Uxx9/rEGDBsnf31/333+/3n33XZvuS2IQAADARpkyKUPZLwRtazsAAADIX+wR691KnBcdHa3o6Ohsz82aNcuq7KWXXtJLL71k833+jcQgAACAjTKNq4c92gEAAED+Yo9Yr6DEeWw+AgAAAAAAABRBjBgEAACwUYadphLbow0AAADYlz1ivYIS55EYBAAAsBGJQQAAgMKrKCUGmUoMAAAAAAAAFEGMGAQAALBRpmFSpmGHXYnt0AYAAADsyx6xXkGJ80gMAgAA2IipxAAAAIUXU4kBAAAAAAAAFGqMGAQAALBRhpyUYYfPVzPs0BcAAADYlz1ivYIS55EYBAAAsJFhpzUGjQKy9gwAAEBRYo9Yr6DEeUwlBgAAKCDWrl2r9u3bKywsTCaTST/88IPFeZPJlO3x3nvvmes0b97c6nznzp3v8JMAAAAgP2DEIAAAgI0ctflIUlKSatWqpV69eqlDhw5W5+Pi4ixe//LLL+rdu7dV3b59+2r06NHm156enjb1AwAAoDArSpuPkBgEAACwUYbhpAzDDmsMGrbVj4qKUlRUVI7nQ0JCLF7/+OOPatGihcqVK2dR7uXlZVUXAAAAV9kj1rM1znMUphIDAAA4WGJiosWRkpJy222ePHlSS5YsUe/eva3Off311woKClK1atU0ePBgXbx48bbvBwAAgIKHEYMAAAA2ypRJmXb4fDVTVz9KDg8PtygfOXKkRo0adVttz549Wz4+Pnr88cctyp966ilFRkYqJCREu3bt0rBhw7Rjxw6tWLHitu4HAABQWNgj1suK8/I7EoMAAAA2svcag7GxsfL19TWXu7u733bbM2bM0FNPPSUPDw+L8r59+5r/Xb16dVWsWFH169fX1q1bVbdu3du+LwAAQEFXlNYYZCoxAACAg/n6+loct5sYXLdunfbu3as+ffrctG7dunXl6uqq/fv339Y9AQAAUPAwYhAAAMBG9tt8JG+mmEyfPl316tVTrVq1blp39+7dSktLU2hoaJ70BQAAoKCxz+YjTCUGAAAolK6uO3P700NsbePSpUs6cOCA+fXhw4e1fft2BQQEqEyZMpKubmTy7bff6v3337e6/uDBg/r666/10EMPKSgoSHv27NGgQYNUp04dNW3a9PYeBgAAoJCwR6xnj1jxTiAxCAAAUEBs3rxZLVq0ML8eOHCgJKlHjx6aNWuWJGn+/PkyDENdunSxut7NzU2///67PvjgA126dEnh4eFq27atRo4cKWdn5zvyDAAAAMg/SAwCAADYKFNOyrDjrsS51bx5cxk3mZby7LPP6tlnn832XHh4uNasWWPTPQEAAIoae8R67EoMAABQSOX3NQYBAABw64rSGoPsSgwAAAAAAAAUQYwYBAAAsFGmnJTpgKnEAAAAyHv2iPUKSpxHYhAAAMBGGYZJGcbt7zRnjzYAAABgX/aI9QpKnMdUYgAAAAAAAKAIYsQgAACAjTLstCtxRgGZYgIAAFCU2CPWKyhxHolBAAAAG2UaTsq0w67EmQVktzoAAICixB6xXkGJ85hKDAAAAAAAABRBjBgEAACwEVOJAQAACi+mEgMAACBHmbLPTnOZt98VAAAA2Jk9Yr2CEucxlRgAAAAAAAAoghgxCAAAYKNMOSnTDp+v2qMNAAAA2Jc9Yr2CEueRGAQAALBRhuGkDDvsSmyPNgAAAGBf9oj1CkqcVzB6CQAAAAAAAMCuSAwCAADYKFMmux0AAADIXxwV502ZMkWRkZHy8PBQvXr1tG7duhzr9uzZUyaTyeqoVq2aTfdkKjGQj7nvSZLPj2fkduiKnM+l68yQMrrSwPfqyXRDfvNOymPbRbmcTJXh5azkGt463y1YmQGu1xpJy5T/nHh5/XFBptRMpdTw1rm+YcoIdM3+psjXqtU6ow5dD6hC5fMKDErR28Ma6M91oebz/sWT1ev5ParT4JSKeadr945AfTqphk4c83Zgr5FXnnr5mLq9fNyi7OxpVz3VsK6DelR0MJUYwO36eVagvp1aUmdPuapspWT1G31cNRom5Vh/5aLi+mZKSZ045K5ivhmq1zxRz444Id+ADEnSH0v9NP/DYJ044q70NKlUZKo69Dullk+cu1OPBBu0i9qrJx7bo4DiV3Q0xl+fTq+v3XtKZls3oPhl9e21VRUrJCgs9KJ+XFxFn02vb1GnaaMYdeq4S2EhF+XikqnjJ3y16Meq+n11uTvxOLiJtl1OqEPvYwookaqjB4pp2thy2r3FL8f61e8+r76vHVbZCklKOOWuhV+U1tIF12L+Jq3OqNNzsQotc0UuLoaOH/XU9zNLaeVPweY6T714VE+9GGPR7tnTrup2byP7P2Ah5IipxAsWLNCAAQM0ZcoUNW3aVJ999pmioqK0Z88elSlTxqr+Bx98oHfeecf8Oj09XbVq1VLHjh1tui+JQSAfMyVnKjXCQ0kt/BU0IdbyXEqm3A5fUeITJZVW1kNOSRnynxmnEu8c1cnxFcz1is+Ml8fmRCW8Eq5Mb2f5z4lX0LijOvluecmZkSoFjYdnhg4f8NNvS8po+NhN15019Ma4/ykj3Ulvv9ZQl5Nc9Fjng/rP5PXq1+1+pSTzX35hdGSvp15/uor5dWYmP9cAkN+t/tFfn44spRfHHlO1Bkla8mWQ3niqnD5f/Y9Klk6zqr9rYzG917+Mnht1XI1aJ+pMnKs+fK20Jg0O18gZRyRJPv4Z6vLySYVXSJaLq6GNv/nq/VfKyD8oXfWbX7zDT4gbue+eI3qu9xZ98tnd2v13ST3UZr/GjFipZ19sr9NnilnVd3XN1IVEd837toYee/jvbNu8eMlN87+trthjfkpPd1KD+sc1sP8Gnb/goS3bwvL6kXAD90Wd1rPDDmnK6Aras9VXUZ3iNHraLvVrV0+n4zys6geXStboz3Zr2bchmvBqZd1VN1HRIw7owjlX/ffXIEnSxQsumv9puI4d8lJamkkNm5/VK2P36fxZN239o7i5rSP7vDT8mRrm1xkZef+8uHUTJ05U79691adPH0nS5MmTtXz5ck2dOlXjxo2zqu/n5yc/v2sJ5h9++EHnzp1Tr169bLovH1PbqGfPnnr00UetylevXi2TyaTz589bnatcubLc3Nx0/Phxi7o3OmbNmnXDevHx8Tn2ceHChWrYsKH8/Pzk4+OjatWqadCgQebzs2bNsmgrNDRUTz75pA4fPmyuExERke19s7LRR44cybFvf/75p7md1NRUjR8/XrVq1ZKXl5eCgoLUtGlTzZw5U2lpabf8nhYVyXV9lNglWFcaWX+aZBRz1ukRkbrSxE/ppdyVWslL53uHyu1QspxPp0qSTEkZKrbynM73CFVKTW+llfNUQv/Sco1JlsfOS3f6cWAHW/4M1pefV9X6tdYBXlh4kqpWP6dP3q+p/f8U1/FYH015v5Y8PNPVrOXxbFpDYZCRYdK5M27m48JZRgPfCRlystuB/IM4jzjvTlk0rYTadDmrqKfOqkzFFD0/+rhKhKVp8ZygbOv/vdVLweGperTPGYWUSVX1hklq2y1B+3Z4mevUanJJTaMuqEzFFIVFpOqxPmdUruoV7f6fdaIJjvX4I39r+W/ltWxFRcUe89Nn0+vr9BkvtYval239k6e89ekXd+v3VeV0OSn73/N/7QrR+j/LKPaYn+LiffTj4io6fMRf1aqeystHQS481vO4fl0YrOXfhSj2kJemjSuv0/HuatslLtv6D3WO06k4d00bV16xh7y0/LsQrVgUrMefOWaus/N//trwW5BiD3kpPtZTP35ZSof3FlO1uhcs2ro+Tkw855anz1qY2DPOS0xMtDhSUlKs7peamqotW7aodevWFuWtW7fW+vXrc9Xn6dOnq2XLlipbtqxNz8rwkTz2xx9/KDk5WR07dtSsWbM0fPhwNWnSRHFx1/4TePnll5WYmKiZM2eay/z8/LRx40ZJ0t69e+Xr62vRbsmS2Q8z/+2339S5c2eNHTtWDz/8sEwmk/bs2aPff//dop6vr6/27t0rwzD0zz//6LnnntPDDz+s7du3y9nZWZI0evRo9e3b1+I6Hx8fq/tdP389MDBQ0tVv7DZt2mjHjh16++231bRpU/n6+urPP//UhAkTVKdOHdWuXftmbyFsYLqcKcMkZRa7+jV0O3RFpnRDybWuTSPNDHBVWriH3PZeVnJtn5yaQgHk6popSUpNcTaXZWaalJ7mpGo1E/TrYtt+QaBgKBWRrK82bFVaqpP2bi+mWRPCFR9r/ekz7CvTMCnTuP3RmfZoA45DnEecdyvSUk3a/5eXOr1ombCp1+yi9mzOPol3V/0kzX43VP/73Ud3339R58+4aN0SfzVomZhtfcOQtv/hrdiD7npmOB8G5ycuLhmqWP6svllo+bO1dXuoqlY5bae7GKpdM16lSyVq+myWF3EkF9dMVah2Ud98XtqifNt/i6tqnex/fqvWTtS2/xa3KNvyR3G17nBSzi6Zyki//kNFQ7UanVfpyCua+b7lgJJSZa/oy7UblZZq0t4dPpo9KULxxzxv+7mKAnvEelnXh4eHW5SPHDlSo0aNsig7c+aMMjIyFBwcbFEeHBx8ww8Ms8TFxemXX37R3Llzbe4nicE8Nn36dHXt2lXNmjXTCy+8oNdff11ubm4KCQkx1/H09FRKSopF2b+VLFlS/v7+ubrf4sWLdc899+jVV181l1WqVMnqk1qTyWS+X2hoqEaOHKlu3brpwIEDqly5sqSrwWFOfcoSGBiYY53Jkydr7dq12rx5s+rUqWMuL1eunDp27KjU1NRcPRNyKTVT/l/F6/I9fjK8rgb9zufTZbiYZHg7W1TN9HOW8/l0R/QSeejYUW+djPNUz3579PF7tZR8xUWPdT6ggKAUFQ9MdnT3kAf2bvfWhMHldfywh/yD0tTlheN6/7s96temhi6eZ+QgkNeI84jzbkXiWWdlZpjkH2Q5Zdi/RJrOncr+Q9tqd1/W0I+Pamy/CKWmOCkj3aRGrS/ohTHHLOolJTqpa91qSkt1kpOzoZfGHlO9ZiQG8xNf3xQ5Oxs6d94yOXPuvKcCip+4rba9vFL19YxFcnXNUGamSR9/2kDbdoTe/ELkGd/iaXJ2kc4nWI7UO5fgquJB1ssGSFLxEmk694dlHHc+wU0uroZ8i6fr3OmrbXl5p+vLNRvl6mYoM1P65K0K2rb+WkJx7w4fvf9aZR0/4in/wFR1fj5WE+bt0PPt6xEn3mGxsbEWHwK6u7vnWNdkskxGGoZhVZadWbNmyd/fP9tR+jdDYjAPXbx4Ud9++602btyoKlWqKCkpSatXr1aLFi3y7J4hISGaO3eudu3aperVq+f6Ok/Pq7+YsqZ92MPXX3+tli1bWgSLWVxdXeXqenv/GaWkpFgMwU1MzP4TlyIh3VDgpFjJkM71ZQ2Roiojw0lj32igl1/bpgW//KKMdJO2bymhTRuyH3mCgm/zGv9rL/ZKf2/11ozVO9Sywxl9P50/BPJSpp2mAWcylbjAIs4jzrtd1/+dZxgm5bSB5dF97pryZmk99Uq86jW/qLOnXPXF22H6cGi4Bk68tg61p3empqzYq+QkZ237w1ufvVVKIWVTVasJycF8x7B8aTIZV78HbsOVK66KHtBWnp5pql0zXs8+s0XxJ731164bfwiAvGdc//XOpszyAsvvBVPWN8y/rrmS5KwXH6srT68M1Wp8Xn1fO6T4Yx7a+T9/SdLmdQH/aqGY/t7uq+m/blLLR0/q+1mWIxhhzR6xXlac5+vrazU74HpBQUFydna2Gh146tQpq1GE1zMMQzNmzNDTTz8tNzfbp4uTGLwFixcvlre35Q6fGdms4jl//nxVrFjRPAWjc+fOmj59us0BY+nSlj+0pUqV0t69e7Ot+9JLL2ndunWqUaOGypYtq0aNGql169Z66qmncsxKHzt2TO+9955Kly6tSpUqmcuHDh2qN954w6Lu4sWL1bx5c/PrJk2ayMnJ8oflwoULcnZ21v79+y3q3khu39N/GzdunN56661ctV+opRsKnBgjl1NpOj0qwjxaUJIy/F1kSjdkupRhMWrQ6UKGMirz418YHdjrr5d6tZBXsTS5uGYq8by7Jk5bo/3/+Du6a7gDUq4468heT5WKYIRoXss0nJRphx2F7dEG7Is4r7n5NXFe3vANyJCTs6Fzpy2TpxfOuKh4iexndCz4KFjV7k5Sx+irU03L3ZUsD89jGvRYRfUYGqfA4KvXOTld3Y1YkspXv6LY/R5a8FFJEoP5SGKiuzIyTCpe/IpFub9fss6dv72lQAzDpLj4q6NODx0OUJnwC+r0xG4Sgw6UeM5VGelS8SDLUdT+gWk6n5D9ByjnTrta1fcLTFN6mkmJ56/9DWcYJsXFXP3g59A/3ipT7rKefDbWnBi8XsoVZx3dV0xhZa9kex6W7BHr2XK9m5ub6tWrpxUrVuixxx4zl69YsUKPPPLIDa9ds2aNDhw4oN69e99SP8kM3IIWLVpo6tSpFmUbN25Ut27dLMqmT59uUdatWzfdd999On/+fK6njEjSunXrLNZ8cXHJ+ctWrFgxLVmyRAcPHtSqVav0559/atCgQfrggw+0YcMGeXldXaD4woUL8vb2lmEYunz5surWratFixZZZJdfffVV9ezZ06L9UqVKWbxesGCBqlatalGWtXZNboe8Srl/T/9t2LBhGjhwoPl1YmKi1dz9Qu//k4Kucak6NSpSmT6W3xup5TxluJjk8dclXWlydb0Jp3Npco1N1oWnb/ypAwq2rIWpw0pfUoXK5/Xl51VvcgUKA1e3TJUpf0W7N7F+KHCriPOuIc7LG65uhirWvKyta33UNOraRgFb1/qocZsL2V6TfMVJzs6Ww4ucnK1HEF3PMKS0VD6AyE/S0521/2CA6tSK1/o/y5jL69SO158b7TuKyyTJ1YVtaB0pPc1JB3b7qE6T89rw27XNheo0Oac/VwZme83f233VsEWCRVndpue0f7d3NusL/ovp6v8vOXFxzVR4+cvateXGI9fgOAMHDtTTTz+t+vXrq3Hjxpo2bZpiYmLUr18/SVd/Nx4/flxz5syxuG769Olq2LChTbMJ/o3E4C0oVqyYKlSoYFF27Jjl+h579uzRxo0btWnTJg0dOtRcnpGRoXnz5un555/P9f0iIyNtCjAlqXz58ipfvrz69Omj4cOHq1KlSlqwYIF522ofHx9t3bpVTk5OCg4OVrFi1gsdBwUFWT3n9cLDw3OsU6lSJf3999+56m9u3tPrubu733BufmFgupIhl/hrnxY5n0yV6+EryvR2VkaAq4ImxMj18BWdGVZWyjTkdO7qFKFMb2fJ1UlGMWcl3V9c/rPjlOntrExvZ/nPiVdaGQ8l1/DO6bbIxzw80xVWKsn8OiT0sspVuKCLF111+qSX7mlxXBfOu+v0SU9FlEvUsy/v1J/rQrVtE9OJC6M+w45q4+/FdeqEm/wD09XlxePy8s7QbwtLOLprhV6GTMrIac6fje0gfyHOu4Y4L+88/uxpvde/jCrVvKyq9ZO09KtAnTruqrbdz0iSZowN1Zl4Vw35MEaS1KhVoia/Gq6fZ19S/eYXdfakqz4dWUqV6yQpMOTqaMH5H5VUxZqXFRaRqrRUkzat9NVv3wXopXGxOfYDjrHox6p6dcB67T8QoL/3llBUm/0qGZSkJcsqSpJ6Pb1NgYGXNWFyU/M15SLPSroaC/r5Jatc5FmlpzspJtZfktSpwy7tOxCouHhvubhkqkG943qgxSF9/GmDO/58sPT9rFIa9O5e7d/lrX+2++rBJ+NUIjRFS+dfXfal58DDCiyZqvdfu7oG7NL5oWr/1An1fe2Qln0Toiq1E9W6w0mNH1zF3OaTz8Zq/y5vxcV4yMXV0N3NzuqBR07pk7eu/V/be8ghbVwVoNMnPMxrDHp5Z+j3Hxggkhv2iPVsvb5Tp05KSEjQ6NGjFRcXp+rVq2vp0qXmXYbj4uIUExNjcc2FCxe0cOFCffDBB7fcTxKDeWT69Om677779Mknn1iUf/nll5o+fbpNAePtioiIkJeXl5KSriUTnJycbhoM3q6uXbvq9ddf17Zt26zWn0lPT1dKSkq2gSqucTt4RSVHHTG/Lj776noDSc39deHJkvLcfFGSFDL4oMV1p0ZFKKX61cTfuZ4h8neWAifGypSaqZQa3jrzYinJmT9GC6KKVc7rnY/+a37dt/8uSdJvS8M1aWxdFQ9MVp8Xd8k/IEXnEjz0+7JwzZ9V2VHdRR4LCknV0A8OyLd4ui6cddE/2731SodqOnWi8P4xnV8wlbhoI84jzrtdzR85r4vnnPX1pBCdPeWispWTNearQwouffVD3rOnXHX6+LURnq07ndWVS076aWaQPn+rlIr5Zah204vqPfzaDtjJl5308evhOhPnKjePTIWXT9GQj46q+SPn7/Tj4SbW/hEhX58UPdVpp4oHXNHRo/56c3QLnTp9NX4PKH5FJYOSLK6ZMnmp+d+VKpzV/c2O6OTJYurx7NUphx4e6Xqx3/8UFHhZqanOij3uq/GTmmrtHxF37LmQvbW/lJCPf5q6vhCjgBKpOrK/mEY+V12nTlydOl68RKpKhF1bU/XkcQ+NeK6ann3tkNp1PaGEU2767D/l9d9fr4049PDMUPSIAwoKSVVqspNiD3tqwpDKWvvLtQ+Hg4JTNPT9vfL1T9OFc67au8NHr3SqZb4vbuxOTyXOEh0drejo6GzPzZo1y6rMz89Ply9ftvk+/0ZiMA+kpaXpyy+/1OjRo62Gcvbp00fjx4/Xjh07VKtWrVy1d+rUKSUnW64XFRgYmO2izqNGjdLly5f10EMPqWzZsjp//rw+/PBDpaWlqVWrVjY9x8WLF60WvvTy8rJYNDMhIcGqjr+/vzw8PDRgwAAtWbJEDzzwgN5++23dc8898vHx0ebNm/Xuu+9q+vTpql27tk19KmpSqnsr9ruchwPf6JyZm5PO9w7T+d5sSlIY7NwWpLb35LzGxM/fldfP35W/gz2CI73zckVHdwEocojziPPspX3PBLXvmZDtucGTY6zKHul9Ro/0PpNjez2Hxqvn0PgczyN/WfxLZS3+JfsPb9//sIlV2YOP5Dz1XpJmf11bs7+ubY+uIQ8smRemJfOy/3ts0jDr74Ndm/zVv0PdHNub80GE5nwQccN7vjuIpYSQO3xMnQd++uknJSQkWCwYmaVixYqqUaOGpk+fnuv2KleurNDQUItjy5Yt2dZt1qyZDh06pO7du6tKlSqKiopSfHy8fv31V1WubNuooREjRljdd8iQIRZ1WrZsaVXnhx9+kHR1CsiKFSs0ZMgQffbZZ2rUqJHuvvtuffjhh+rfv/8tz38HAMDRMnRtisntHShoiPN+kEScBwAo3OwT6xUMJsO44SbZQIGRmJgoPz8/VfjyNTl7MTy6KIocwo57RV3mCUZKFFXpRqpWJn+jCxcuWIx4sres3zVv/NlaHt7Z7yZoi+RLaRrT6Nc87zdQ0GX97J3bV06+PoxtKIoefORpR3cBDuS096ijuwAHSjdS9XviV3ckXrJnrFdQ4jx+qwIAAAAAAABFEGsMAgAA2CjDcFKGHTYOsUcbAAAAsC97xHoFJc4jMQgAAGAjQyZl6vZ3dzfs0AYAAADsyx6xXkGJ8wpG+hIAAAAAAACAXTFiEAAAwEZMJQYAACi8mEoMAACAHGUaJmUatz89xB5tAAAAwL7sEesVlDivYKQvAQAAAAAAANgVIwYBAABslCEnZdjh81V7tAEAAAD7skesV1DiPBKDAAAANmIqMQAAQOHFVGIAAAAAAAAAhRojBgEAAGyUKSdl2uHzVXu0AQAAAPuyR6xXUOK8gtFLAACAfCTDMNntsMXatWvVvn17hYWFyWQy6YcffrA437NnT5lMJoujUaNGFnVSUlL00ksvKSgoSMWKFdPDDz+sY8eO3e5bAgAAUGg4Is5zFBKDAAAABURSUpJq1aqljz/+OMc6Dz74oOLi4szH0qVLLc4PGDBA33//vebPn68//vhDly5dUrt27ZSRkZHX3QcAAEA+w1RiAAAAG9l785HExESLcnd3d7m7u1vVj4qKUlRU1A3bdHd3V0hISLbnLly4oOnTp+vLL79Uy5YtJUlfffWVwsPD9dtvv6lNmza38hgAAACFCpuPAAAAIEeG4aRMOxyGcTUUCw8Pl5+fn/kYN27cLfdt9erVKlmypCpVqqS+ffvq1KlT5nNbtmxRWlqaWrdubS4LCwtT9erVtX79+lt/QwAAAAoRe8R6WXFefseIQQAAAAeLjY2Vr6+v+XV2owVzIyoqSh07dlTZsmV1+PBhvfnmm7r//vu1ZcsWubu7Kz4+Xm5ubipevLjFdcHBwYqPj7+tZwAAAEDBQ2IQAADARhkyKUO3Pz0kqw1fX1+LxOCt6tSpk/nf1atXV/369VW2bFktWbJEjz/+eI7XGYYhk6lgTHcBAADIa/aI9ewRK94JBWNcIwAAQD6SaVxbe+b2jrztZ2hoqMqWLav9+/dLkkJCQpSamqpz585Z1Dt16pSCg4PztjMAAAAFhH1iPUc/Re6QGAQAACikEhISFBsbq9DQUElSvXr15OrqqhUrVpjrxMXFadeuXWrSpImjugkAAAAHYSoxAACAjbIWlbZHO7a4dOmSDhw4YH59+PBhbd++XQEBAQoICNCoUaPUoUMHhYaG6siRI3r99dcVFBSkxx57TJLk5+en3r17a9CgQQoMDFRAQIAGDx6sGjVqmHcpBgAAKOrsEevZI1a8E0gMAgAA2ChTJmXaYd0YW9vYvHmzWrRoYX49cOBASVKPHj00depU7dy5U3PmzNH58+cVGhqqFi1aaMGCBfLx8TFfM2nSJLm4uOjJJ5/UlStX9MADD2jWrFlydna+7ecBAAAoDOwR69kjVrwTSAwCAAAUEM2bN5dh5LxgzfLly2/ahoeHhz766CN99NFH9uwaAAAACiASgwAAADbKMEzKMOywK7Ed2gAAAIB92SPWKyhxHolBAAAAGzlqjUEAAADkvaK0xmDB6CUAAAAAAAAAu2LEIAAAgI0yZVKmHaaHFJRFqQEAAIoSe8R6BSXOIzEIAABgI8NOuxIbBSRgBAAAKErsEesVlDiPqcQAAAAAAABAEcSIQQAAABtlGnaaSlxAdqsDAAAoSuwR6xWUOI/EIAAAgI3YlRgAAKDwYldiAAAAAAAAAIUaIwYBAABsxFRiAACAwoupxAAAAMhRpp12JbZHGwAAALAve8R6BSXOYyoxAAAAAAAA4GBTpkxRZGSkPDw8VK9ePa1bt+6G9VNSUjR8+HCVLVtW7u7uKl++vGbMmGHTPRkxCAAAYCOmEgMAABRejphKvGDBAg0YMEBTpkxR06ZN9dlnnykqKkp79uxRmTJlsr3mySef1MmTJzV9+nRVqFBBp06dUnp6uk33JTEIAABgIxKDAAAAhZcjEoMTJ05U79691adPH0nS5MmTtXz5ck2dOlXjxo2zqr9s2TKtWbNGhw4dUkBAgCQpIiLC5n4ylRgAAAAAAADIA4mJiRZHSkqKVZ3U1FRt2bJFrVu3tihv3bq11q9fn227P/30k+rXr6/x48erVKlSqlSpkgYPHqwrV67Y1D9GDAIAANiIEYMAAACFlz1HDIaHh1uUjxw5UqNGjbIoO3PmjDIyMhQcHGxRHhwcrPj4+GzbP3TokP744w95eHjo+++/15kzZxQdHa2zZ8/atM4giUEAAAAAAAAgD8TGxsrX19f82t3dPce6JpNlMtIwDKuyLJmZmTKZTPr666/l5+cn6ep05CeeeEKffPKJPD09c9U/EoMAAAA2YsQgAABA4WXPEYO+vr4WicHsBAUFydnZ2Wp04KlTp6xGEWYJDQ1VqVKlzElBSapataoMw9CxY8dUsWLFXPWTNQYBAABsZEjKlOm2D8PRDwIAAAAr9oj1bInz3NzcVK9ePa1YscKifMWKFWrSpEm21zRt2lQnTpzQpUuXzGX79u2Tk5OTSpcunet7kxgEAAAAAAAAHGjgwIH64osvNGPGDP3999965ZVXFBMTo379+kmShg0bpu7du5vrd+3aVYGBgerVq5f27NmjtWvX6tVXX9UzzzyT62nEElOJAQAAbMZUYgAAgMLLnlOJc6tTp05KSEjQ6NGjFRcXp+rVq2vp0qUqW7asJCkuLk4xMTHm+t7e3lqxYoVeeukl1a9fX4GBgXryySc1ZswYm+5LYhAAAMBGJAYBAAAKL0ckBiUpOjpa0dHR2Z6bNWuWVVmVKlWsph/biqnEAAAAAAAAQBHEiEEAAAAbMWIQAACg8HLUiEFHIDEIAABgIxKDAAAAhVdRSgwylRgAAAAAAAAoghgxCAAAYCPDMMmww6fA9mgDAAAA9mWPWK+gxHkkBgEAAGyUKZMyZYepxHZoAwAAAPZlj1ivoMR5TCUGAAAAAAAAiiBGDAIAANiIzUcAAAAKr6K0+QiJQQAAABuxxiAAAEDhVZTWGGQqMQAAAAAAAFAEMWIQAADARkwlBgAAKLyYSgwAAIAcMZUYAACg8GIqMQAAAAAAAIBCjRGDKHS8fvORs5uHo7sBB1jyx1eO7gIcrE1YbUd3AQ6SaaTd0fsZdppKXFA+SQbyiwfefIY4r4gav+BTR3cBDvSfcrUd3QU4UMYdjvMk+8R6BSXOIzEIAABgI0OSYdinHQAAAOQv9oj1Ckqcx1RiAAAAAAAAoAgiMQgAAGCjTJnsdthi7dq1at++vcLCwmQymfTDDz+Yz6WlpWno0KGqUaOGihUrprCwMHXv3l0nTpywaKN58+YymUwWR+fOne3xtgAAABQKjojzHIXEIAAAgI2ydqqzx2GLpKQk1apVSx9//LHVucuXL2vr1q168803tXXrVi1atEj79u3Tww8/bFW3b9++iouLMx+fffbZLb8XAAAAhY0j4jxHYY1BAACAAiIqKkpRUVHZnvPz89OKFSssyj766CM1aNBAMTExKlOmjLncy8tLISEhedpXAAAA5H+MGAQAALBR5v/vVGePQ5ISExMtjpSUFLv088KFCzKZTPL397co//rrrxUUFKRq1app8ODBunjxol3uBwAAUBjYM87L7xgxCAAAYCPDsNOuxP/fRnh4uEX5yJEjNWrUqNtqOzk5Wa+99pq6du0qX19fc/lTTz2lyMhIhYSEaNeuXRo2bJh27NhhNdoQAACgqLJHrGePWPFOIDEIAADgYLGxsRbJO3d399tqLy0tTZ07d1ZmZqamTJlica5v377mf1evXl0VK1ZU/fr1tXXrVtWtW/e27gsAAICChcQgAACAjey1oHRWG76+vhaJwduRlpamJ598UocPH9bKlStv2m7dunXl6uqq/fv3kxgEAACQfWI9Nh8BAAAopOydGLSXrKTg/v37tWrVKgUGBt70mt27dystLU2hoaF27QsAAEBBRWIQAAAA+c6lS5d04MAB8+vDhw9r+/btCggIUFhYmJ544glt3bpVixcvVkZGhuLj4yVJAQEBcnNz08GDB/X111/roYceUlBQkPbs2aNBgwapTp06atq0qaMeCwAAAA5CYhAAAMBGmYZJJjt8CmzrbnWbN29WixYtzK8HDhwoSerRo4dGjRqln376SZJUu3Zti+tWrVql5s2by83NTb///rs++OADXbp0SeHh4Wrbtq1GjhwpZ2fn23sYAACAQsIesR67EgMAABRS9t6VOLeaN28u4wYX3eicdHX34zVr1th2UwAAgCKmKO1K7OToDgAAAAAAAAC48xgxCAAAYKOrnyLbY/MRO3QGAAAAdmWPWK+gxHkkBgEAAGyUX3clBgAAwO0rSrsSM5UYAAAAAAAAKIIYMQgAAGAj4/8Pe7QDAACA/MUesV5BifNIDAIAANiIqcQAAACFF1OJAQAAAAAAABRqjBgEAACwFXOJAQAACq8iNJeYEYMAAAC2+v/pJbd7qIBMMQEAAChSHBTnTZkyRZGRkfLw8FC9evW0bt26HOuuXr1aJpPJ6vjnn39suieJQQAAAAAAAMCBFixYoAEDBmj48OHatm2b7r33XkVFRSkmJuaG1+3du1dxcXHmo2LFijbdl8QgAACAjQzDfgcAAADyF0fEeRMnTlTv3r3Vp08fVa1aVZMnT1Z4eLimTp16w+tKliypkJAQ8+Hs7GzTfUkMAgAA2Mge04jttbMxAAAA7MuecV5iYqLFkZKSYnW/1NRUbdmyRa1bt7Yob926tdavX3/DvtapU0ehoaF64IEHtGrVKpuflcQgAAAAAAAAkAfCw8Pl5+dnPsaNG2dV58yZM8rIyFBwcLBFeXBwsOLj47NtNzQ0VNOmTdPChQu1aNEiVa5cWQ888IDWrl1rU//YlRgAAMBW9to4hBGDAAAA+Y89Yr3/vz42Nla+vr7mYnd39xwvMZks72kYhlVZlsqVK6ty5crm140bN1ZsbKwmTJig++67L9fdJDEIAABgI3utD8gagwAAAPmPPWK9rOt9fX0tEoPZCQoKkrOzs9XowFOnTlmNIryRRo0a6auvvrKpn0wlBgAAAAAAABzEzc1N9erV04oVKyzKV6xYoSZNmuS6nW3btik0NNSmezNiEAAAwFbG/x/2aAcAAAD5iz1iPRuvHzhwoJ5++mnVr19fjRs31rRp0xQTE6N+/fpJkoYNG6bjx49rzpw5kqTJkycrIiJC1apVU2pqqr766istXLhQCxcutOm+JAYBAABsZK8dhdmVGAAAIP+xR6xn6/WdOnVSQkKCRo8erbi4OFWvXl1Lly5V2bJlJUlxcXGKiYkx109NTdXgwYN1/PhxeXp6qlq1alqyZIkeeughm+5LYhAAAAAAAABwsOjoaEVHR2d7btasWRavhwwZoiFDhtz2PXOVGPzwww9z3WD//v1vuTMAAAAFRiGZBkycBwAAkI1CEuvdTK4Sg5MmTcpVYyaTiYARAAAUeoVpKjFxHgAAgCVHTCV2lFwlBg8fPpzX/QAAAIADEOcBAAAUXU63emFqaqr27t2r9PR0e/YHAAAg/zPseORDxHkAAKBIK8Rx3vVsTgxevnxZvXv3lpeXl6pVq2beEaV///5655137N5BAACA/MdkxyP/IM4DAACQCmOclxObE4PDhg3Tjh07tHr1anl4eJjLW7ZsqQULFti1cwAAALhziPMAAACKllytMfhvP/zwgxYsWKBGjRrJZLqW/bzrrrt08OBBu3YOAAAgX7LX9JB8NsWEOA8AAED2ifXyWZyXE5sTg6dPn1bJkiWtypOSkiwCSAAAgEKrkCYGifMAAABUpBKDNk8lvvvuu7VkyRLz66wg8fPPP1fjxo3t1zMAAADcUcR5AAAARYvNIwbHjRunBx98UHv27FF6ero++OAD7d69Wxs2bNCaNWvyoo8AAAD5i2G6etijnXyEOA8AAED2ifXyWZyXE5tHDDZp0kT//e9/dfnyZZUvX16//vqrgoODtWHDBtWrVy8v+ggAAJCvGIb9jvyEOA8AAKBwxnk5sXnEoCTVqFFDs2fPtndfAAAA4GDEeQAAAEXHLSUGMzIy9P333+vvv/+WyWRS1apV9cgjj8jF5ZaaAwAAKFgK6eYjEnEeAABAUdp8xOYIb9euXXrkkUcUHx+vypUrS5L27dunEiVK6KefflKNGjXs3kkAAIB8pZCuMUicBwAAINYYvJE+ffqoWrVqOnbsmLZu3aqtW7cqNjZWNWvW1LPPPpsXfQQAAMAdQJwHAABQtNg8YnDHjh3avHmzihcvbi4rXry4/vOf/+juu++2a+cAAADyI5Nx9bBHO/kJcR4AAIB9Yr38FuflxOYRg5UrV9bJkyetyk+dOqUKFSrYpVMAAAD5mmHHIx8hzgMAAFChjPNykqvEYGJiovkYO3as+vfvr++++07Hjh3TsWPH9N1332nAgAF6991387q/AAAAsCPiPAAAgKIrV1OJ/f39ZTJdWzTRMAw9+eST5jLDuJoGbd++vTIyMvKgmwAAAPlIIdp8hDgPAADgOkVo85FcJQZXrVqV1/0AAAAoOOw1PSQfTDEhzgMAALiOPWK9fBDn5UauEoPNmjXL634AAADgJtauXav33ntPW7ZsUVxcnL7//ns9+uij5vOGYeitt97StGnTdO7cOTVs2FCffPKJqlWrZq6TkpKiwYMHa968ebpy5YoeeOABTZkyRaVLl3bAEwEAAMCRbN6VOMvly5cVExOj1NRUi/KaNWvedqcAAADyNQeNGExKSlKtWrXUq1cvdejQwer8+PHjNXHiRM2aNUuVKlXSmDFj1KpVK+3du1c+Pj6SpAEDBujnn3/W/PnzFRgYqEGDBqldu3basmWLnJ2dJRHnAQCAIo4Rgzk7ffq0evXqpV9++SXb86w9AwAACj0HJQajoqIUFRWVfVOGocmTJ2v48OF6/PHHJUmzZ89WcHCw5s6dq+eee04XLlzQ9OnT9eWXX6ply5aSpK+++krh4eH67bffVLduXeI8AACAIpQYzNWuxP82YMAAnTt3Tn/++ac8PT21bNkyzZ49WxUrVtRPP/2UF30EAAAo1P69M3BiYqJSUlJsbuPw4cOKj49X69atzWXu7u5q1qyZ1q9fL0nasmWL0tLSLOqEhYWpevXqWr9+PXEeAABAEWPziMGVK1fqxx9/1N133y0nJyeVLVtWrVq1kq+vr8aNG6e2bdvmRT8BAADyDzvvShweHm5RPHLkSI0aNcqmpuLj4yVJwcHBFuXBwcE6evSouY6bm5uKFy9uVSc+Pp44DwAAQGJX4htJSkpSyZIlJUkBAQE6ffq0KlWqpBo1amjr1q127yAAAEB+YzKuHvZoR5JiY2Pl6+trLnd3d7/1Nk2WQahhGFZl18uqQ5wHAABgn1jPHrHinWBzYrBy5crau3evIiIiVLt2bX322WeKiIjQp59+qtDQ0LzoI4B/KeFzSS+13qgmFWPk4ZKhowl+evuH5vonroQk6dkWm9S6+kEF+11SWoaT/j5RQlN+b6Ddx4Jv0jLyo51/FtO3U0pq/04vnT3pqpHTD6tJ1AXz+XOnXTT9P2HassZHSRecVb3RJb0w5phKlbu2YcAHQ0pr2zofJZx0ladXpqrWT1Lv4SdUpqLtUxWRP7XrcUYdnz+tgJJpOrrPQ5+OCNOu/3k7uluwga+vr0Vi8FaEhIRIujoq8N8x2alTp8yjCENCQpSamqpz585ZjBo8deqUmjRpQpwH3EEdGu/WU812KNDnsg6fLK5JPzXRjiPZ/5w1r35Ijzfao4phCXJzydChk8X1xYr62rjv2mjjtvX26s1Oq62uve/13kpNv+U9J5FHNn8ZqD8/L6lLp1xVolKyWr1xXGUaJGVb9+dXy+ivhQFW5UEVr+i55XslSTu+C9DiIWWs6gz9e4dc3AtIdgLZIs5DXrulNQbj4uIkXZ3msmzZMpUpU0Yffvihxo4da1NbPXv21KOPPmpR9t1338nDw0Pjx4+XJI0aNUomk8nqqFKlivma5s2ba8CAARavTSaT5s+fb9H25MmTFRERYX49a9asbNv28PDIsc+rV6+WyWTS+fPnrc5FRERo8uTJVuVjx46Vs7Oz3nnnHYu62d0762jevPkN6/27resdOnRIXbp0UVhYmDw8PFS6dGk98sgj2rdvn7nOv9vy8fFR/fr1tWjRIvP53L7v2dXp16+fRX9WrVqlhx56SIGBgfLy8tJdd92lQYMG6fjx47f8nhZVPh4pmt7nB6VnOOnlLx9Sx4+f1OTljXUx2c1c5+gZf41fco86f/Kk+nzxqOLO++iT7kvk73XFgT3HrUq+7KRy1a7ohf8cszpnGNJbz0Qq7qibRs08pE9+3avg0ql6rVMFJV++9t97xZpXNGhSjD5f84/+M/egZEivdykv9hAoHJo9fE793jqheR+WVHTrStq1sZjGfH1YJUql3vxi3DrDjoedREZGKiQkRCtWrDCXpaamas2aNWrSpIkkqV69enJ1dbWoExcXp127dqlJkybEecR55jrEeXmrZa0DGtB+vWatrKMeH3TQ9sMhmtR7qYL9L2Zbv3ZknP63v7QGzohSzw87aMvBME3ouUyVws5Y1Lt0xU0PjX7a4iApmP/sWeyvFWNKqekLJ9Vn8V6F10/S/GfK6cJx12zrt3rzmF7euMt8vPTf3fL0T1fVf31YLEnu3hkW9V7euIukYAFHnOdA+SzOy0s2/5Z46qmnzP+uU6eOjhw5on/++UdlypRRUFDQbXXmiy++0AsvvKBPPvlEffr0MZdXq1ZNv/32m0VdF5cbd93Dw0NvvPGGOnToIFfX7P+Dla5+Qr93716LsptNt7HVzJkzNWTIEM2YMUOvvfaaJGnTpk3mnf3Wr1+vDh06aO/evebRAm5u1xI9o0ePVt++fS3a9PHxyfZeqampatWqlapUqaJFixYpNDRUx44d09KlS3XhguUvjpkzZ+rBBx/U+fPn9d5776ljx476448/1LhxY0m5e9/79u2r0aNHW5R5eXmZ//3ZZ58pOjpaPXr00MKFCxUREaGYmBjNmTNH77//viZOnHjjNw8Wety7TScTvTX6hxbmsrjzliNMlu+saPF60rImerTeP6oYkqBNh0rfkX7Cfu6+/6Luvj/7PxKOH3LX31uK6bNV/yiicrIk6cVxx9SpZnWt+t5fUU+dlSQ91C3BfE1IuNRjaJyeb1lFJ2PdFBZBUFHQPf7sGS2fF6BlcwMlSZ+OLKV6zS+qXfcEzRzHCK/C5tKlSzpw4ID59eHDh7V9+3YFBASoTJkyGjBggMaOHauKFSuqYsWKGjt2rLy8vNS1a1dJkp+fn3r37q1BgwYpMDBQAQEBGjx4sGrUqKGWLVvK2dnZ3DZxXu4Q5xHn3You9+7Uz5uq6Kf/VZUkTf65qRpVOqbHG+3R1GUNrepP/rmpxetPlzXUfXcd1T13HdW+E9d+Ng1JZy95CfnbxuklVLvjWdXpdDVWaz3iuA6t89HWr4PUYkicVX0P30xJmebXe3/105ULzqrVMcGyoknyLpGel13HHUachzvhtj8+8vLyUt26dW+7I+PHj9eIESM0d+5cdejQweKci4uLeXpMbnXp0kU///yzPv/8c0VHR+dYz2Qy2dy2LdasWaMrV65o9OjRmjNnjtauXav77rtPJUqUMNcJCLg6LLxkyZLy9/e3asPHxyfXfdyzZ48OHTqklStXqmzZspKksmXLqmnTplZ1/f39FRISopCQEH366aeaP3++fvrpJ3PAmJv33cvLK8c6x44dU//+/dW/f39NmjTJXB4REaH77rsv20+OcWP3VT6qPw+U1jtP/qq6ESd0+mIxffu/avphy13Z1ndxztBj9ffo4hU37YsPvMO9RV5LS736x62b+7VA0dlZcnU1tHuTtzkx+G/Jl53064IAhZRJUYmwtDvWV+QNF9dMVax5WQs+LmlRvmWNj+6qn/10JBRsmzdvVosW1z4cGjhwoCSpR48emjVrloYMGaIrV64oOjpa586dU8OGDfXrr79aJJomTZokFxcXPfnkk7py5YoeeOABzZo1yyIpmIU478aI84jzboWLc4YqlzqtOatqW5Rv3F9aNSJO5qoNk8mQl3uaEi9brkfq6Zam74d9LWeToX1xgZq2/G6LxCEcLyPVpLhdXmrc75RFebl7L+rY1mK5amP7NwGKbHpRfqUsY7nUy0766J67ZGRKwVWvqNnAeIVUY9ZQQUWchzslV4nBrKAzN27lk8HXXntNn3zyiRYvXqyWLVvafH12fH199frrr2v06NHq0aOHihXL3X+y9jZ9+nR16dJFrq6u6tKli6ZPn6777rsvz+5XokQJOTk56bvvvtOAAQOyDfKz4+rqKhcXF6Wl2S9R8O233yo1NVVDhgzJ9nx2wbEtUlJSlJJybY20xMTE22qvIChVPFEd7t6jrzfU1My1dVWt9CkNfui/Skt31pIdlc317ql0VGM7rpCHa7rOXPLSC7Pb6cJlTwf2HHkhvEKygkunasa4UL387jF5eGVq0WcldPaUq86etPzv/edZgfpiTJiSLzsrvEKyxs0/KFe3AjK2HTnyDciQs4t0/ozl1/v8aRcVL8mIgbxkkp02H7GxfvPmzWUYOd/YZDJp1KhRN9zR2MPDQx999JE++ugjSVfjvH8ndm6EOM8ScR5x3q3wL5YsF2dDZy9ZxmZnL3oq0Odyrtroet8Oebql6fcd5c1lR077a8w3zXUgPlDF3FPV6Z6dmhb9o56e/IRiz/jZ9Rlw6y6fc5aRYZJ3kOXPY7HANF06nf1o4X+7eMpFB9f46tHJRy3Kg8onq/17MSpZ+YpSLjpr06wSmt2xovou+UcBkcwQKYiI8xzLHrFewdiTOJdrDG7bti1Xx/bt223uwC+//KJ3331XP/74Y47B4s6dO+Xt7W1x/HsKSk6io6Pl4eFxwyD2woULVm23bt36pm2XLl3a6rqYmBiLOomJiVq4cKG6desmSerWrZu+++47mwOboUOHWt1r9erV2dYtVaqUPvzwQ40YMULFixfX/fffr7fffluHDh3Ksf2UlBSNGTNGiYmJeuCBB8zluXnfp0yZYlVn9uzZkqT9+/fL19c314uV5+Y9/bdx48bJz8/PfISHh+dYt7BwMhn6Jy5IU35rqL3xQVq0+S79sKWqOjTYY1Fv8+EwdZ3aUc988Zg27C+jcZ1WqHgxPi0sbFxcpTe/OKzjBz30xF019HD5mtqxwVt3358op+v+Vrz/8XOa8uteTVi0X6UiU/Sf5yKUmlxQflXhZq7PE5lMKjBrmhRYhsl+h4MR51kjzruKOC9vZfd/t5GL/xNa1T6gPq226I2vW+pc0rXk4u6YYC3bVkkH4gK140iohn/dSjFn/NSxyS57dx32cN2X2pBJuVnp4K/vAuThm6HKrSyXDyhV57JqPHpOwVWTVaZBkh7/+IgCIlO0aU6JHFpCQUGc5yCFJM7LjVyNGFy1alWedaBmzZo6c+aMRowYobvvvjvbNVUqV66sn376yaIsp7VX/s3d3V2jR4/Wiy++qOeffz7bOj4+Ptq6datFmafnzUdWrVu3zqoPWQtJZ5k7d67KlSunWrVqSZJq166tcuXKaf78+Xr22Wdveo8sr776qnr27GlRVqpUqRzrv/DCC+revbtWrVqljRs36ttvv9XYsWP1008/qVWrVuZ6Xbp0kbOzs65cuSI/Pz9NmDBBUVFR5vO5ed+feuopDR8+3KKsZMmrQ50Nw7BpHZ/cvKf/NmzYMIvRrImJiYUyaPy3M5e8dPh0cYuyw6eL6/67LP8gSE5z1bGzfjp21k+7jgVr0ctz9UjdvzVr3e1PB0P+UrHmFU39ba+SEp2UlmaSf2CG+retqEo1LUccFPPNVDHfVJUql6oqdY+oQ9Xq+u8vfmrx2HnHdBx2kXjWWRnpUvHr1hPyC0rXudMsNo/cIc6zRpx3FXFe3jif5KH0DJMCfSw/tC3ufcVqFOH1WtY6oOFPrNHrX7XUpgM3XjvaMEz6O7aEwoMu3LAe7iyv4hkyORu6dNpyfdTLCS4qFnTjUWCGIe34NlA1Hj0r55vM/DA5SWE1L+vsEfcb1kP+RZyHO8Xh302lSpXSwoUL1aJFCz344INatmyZVdDg5uamChUq3FL73bp104QJEzRmzBiLneqyODk53VLbkZGRVlMkrl+wecaMGdq9e7dFeWZmpqZPn25TwBgUFGRzH318fPTwww/r4Ycf1pgxY9SmTRuNGTPGImCcNGmSWrZsKV9fX3OQ92+5ed/9/PxyrFOpUiVduHBBcXFxufo0OTfv6b+5u7vL3b1o/aLbEROiskHnLcrKBp5X3Pkb/wFlkuTmwha0hVkx36vrDB4/5Kb9O7zU49X4G19gmJSWavPG9Mhn0tOctP8vL9W976LWL7s2TazufRe1YTnTxvKUvXaaK+Sf+BPn3RxxXvYKc5yXnuGsvcdLqEHFY1qzO9Jc3qDiMa3dHZHjda1qH9Dwjqs1Yu4DWv9P2VzcyVDFsAQdjA+4/U7DbpzdDIVWv6zDf/ioSptrSdvDf/ioUssbJ3FjNnrr3FF31XrSeh3p6xmGdHKPp0pUZtZQQUWc52D2iPUKSJyXL/4qLFOmjNasWaNTp06pdevWdl1DxMnJSWPHjtXUqVN15MgRu7V7Mzt37tTmzZu1evVqbd++3XysXbtWmzZt0q5dd25Iv8lkUpUqVZSUZLlAaUhIiCpUqJBtsGgPTzzxhNzc3DR+/Phsz7Mote3mrq+pGqVPqdd9W1U64ILa1Nivx+r/rW//V02S5OGapuiWG1W99EmF+F1U5dDTeuOR1Srpm6TfdpW/SevIj64kOengLk8d3HV1BEF8rJsO7vLUqWNXP2Ve+7Ofdqz3VtxRN61f5qthnSuo8YMXVK/51Z2M4466af5HJbX/r6vX7Nnspf88FyE3z0w1eKDwrNdUlC2aFqQHu55V684JCq+QrOdGHVfJUmlaMocNh/KUYcejkCPOy1vEeQXTvHU19HCDf9Su/j+KKHlOL7dfr2D/S/r+z6sbyj3/4EaN6LTSXL9V7QMa2WmVPlrcWLuOBivA+7ICvC+rmMe1dRh7t9yshpViFRaQqIqhZzS84xpVCkswt4n8o2Hv09r+TYC2fxOgMwfcteLtMF044aq6T52RJK0aH6qfBpWxum77NwEKq52kkpWTrc6t/SBYB9f66FyMm+L3eGrx0HCd/NtTdZ9KsKqLgoM4z4GKUJzn8BGDWUqXLq3Vq1erRYsWat26tZYvXy4/v6tZ8PT0dMXHW45+MZlMCg4OzlXb7dq1U8OGDfXZZ59ZXWMYhlXb0tVpEk5Ot543nT59uho0aJDtAtSNGzfW9OnTc73Q98WLF6366OXlJV9fX6u627dv18iRI/X000/rrrvukpubm9asWaMZM2Zo6NChNj1Dbt73y5cvW9Vxd3dX8eLFFR4erkmTJunFF19UYmKiunfvroiICB07dkxz5syRt7e33n//fZv6VNTtOVFSg+e10YutNqpPsy06cd5H7//SRMv+qiRJyjRMigg6r3adl8vfK1kXLntoz/GS6jv9ER06zafFBdG+HV4a8sS10Rqfjbo6vazVk2c1eHKMzp501WejSun8GRcFlExXy45n1XXAtR0N3dwztWujt77/vIQuXXCWf1C6ajS6pEk/7pf/TaaroGBY81Nx+RTP0FOvnFRAyXQd3euhN7pF6tRxN0d3DTAjzssZcV7R9NuOCvLzSlHvllsU6HtZh+IDNHBGlOL/fxZIkO9lhfhfMtd/rOEeuThn6tXH/tCrj/1hLl+yuZLe/ubqTuXenql6rcNaBfpc1qVkN+07HqR+U9trT2zeJIdx6+5qd16Xzznrj49CdOm0i0pUSlbnGYfMuwxfOu2qCycsf48nJzrpn2X+aj3iWLZtpiQ6a+nr4Uo64yJ3nwyF3HVFT8/fr1K1crehDfIn4ryiZ8qUKXrvvfcUFxenatWqafLkybr33ntvet1///tfNWvWTNWrV7d5Xeh8kxiUrk43WbNmjVq0aKFWrVrp119/lSTt3r3baoqCu7u7kpOtPynJybvvvqsmTZpYlScmJmY7/SEuLk4hISE2PsFVqamp+uqrr3IM0Dp06KBx48bp3XfflZvbzX+gR4wYoREjRliUPffcc/r000+t6pYuXVoRERF66623dOTIEZlMJvPrV155xabnyM37/vnnn+vzzz+3qNOmTRstW7ZM0tWFwStVqqQJEyboscce05UrVxQREaF27drZtNs1rvljX1n9sS/76SOp6S4aMr/NHe4R8lKtJpe0/MT2HM8/2ueMHu1zJsfzgSHpGvNVzovSo3BYPDtIi2cHObobRYrJsNOuxAXkk2R7IM7LHnFe0bVwQzUt3FAt23NZyb4s0Z89fNP2Pvi5iT742frnAPlT/acTVP/p7EfztX/PemMeD99MDd3zV47ttXrzhFq9ecJu/UP+QZznGPaI9Wy9fsGCBRowYICmTJmipk2b6rPPPlNUVJT27NmjMmWsRxFnuXDhgrp3764HHnhAJ0+ezLFezv00rt/jBiiYEhMT5efnp5o9/iNnNw9HdwcOsHn0VEd3AQ7WJqy2o7sAB0k30rRaP+rChQvZjrSyl6zfNRFj/iMnj9v/XZOZnKwjbwzP834DBV3Wz17dTmOI84qo8aOtk+UoOv5TrrajuwAHulNxnmTfWM/WOK9hw4aqW7eupk699ndt1apV9eijj2rcuHE5Xte5c2dVrFhRzs7O+uGHH2weMXhLcyi+/PJLNW3aVGFhYTp69KgkafLkyfrxxx9vpTkAAADkE8R5AAAA9pOYmGhxpKSkWNVJTU3Vli1b1Lp1a4vy1q1ba/369Tm2PXPmTB08eFAjR4685f7ZnBicOnWqBg4cqIceekjnz59XRsbVnU79/f01efLkW+4IAABAgVFINx8hzgMAAJBd47zw8HD5+fmZj+xG/505c0YZGRlW6yUHBwdnu16yJO3fv1+vvfaavv76a7m43PpKgTYnBj/66CN9/vnnGj58uJydnc3l9evX186dO2+5IwAAAAVF1roz9jjyE+I8AAAA+8Z5sbGxunDhgvkYNmxYzvc1mSxeG4ZhVSZJGRkZ6tq1q9566y1VqlTptp7V5pTi4cOHVadOHatyd3d3JSUl3VZnAAAA4DjEeQAAAPbl6+t70zUGg4KC5OzsbDU68NSpU1ajCCXp4sWL2rx5s7Zt26YXX3xRkpSZmSnDMOTi4qJff/1V999/f676Z/OIwcjIyGwXMvzll19011132docAABAwWOY7HfkI8R5AAAAuuNxnpubm+rVq6cVK1ZYlK9YsUJNmljvOO/r66udO3dq+/bt5qNfv36qXLmytm/froYNG+b63jaPGHz11Vf1wgsvKDk5WYZh6H//+5/mzZuncePG6YsvvrC1OQAAgILHXusD5rOpxMR5AAAAsk+sZ+P1AwcO1NNPP6369eurcePGmjZtmmJiYtSvXz9J0rBhw3T8+HHNmTNHTk5Oql69usX1JUuWlIeHh1X5zdicGOzVq5fS09M1ZMgQXb58WV27dlWpUqX0wQcfqHPnzrY2BwAAgHyCOA8AAMAxOnXqpISEBI0ePVpxcXGqXr26li5dqrJly0qS4uLiFBMTY/f73tK2JX379lXfvn115swZZWZmqmTJkvbuFwAAQL5lr41D8tvmIxJxHgAAgD1ivVu5Pjo6WtHR0dmemzVr1g2vHTVqlEaNGmXzPW99P2NdXRwRAACgyCmkU4n/jTgPAAAUWQ6YSuwoNicGIyMjs90qOcuhQ4duq0MAAABwDOI8AACAosXmxOCAAQMsXqelpWnbtm1atmyZXn31VXv1CwAAIP+y01Ti/PZJMnEeAACA7BPr5bM4Lyc2JwZffvnlbMs/+eQTbd68+bY7BAAAkO8V0qnExHkAAAAqUlOJnezVUFRUlBYuXGiv5gAAAJBPEOcBAAAUTre1+ci/fffddwoICLBXcwAAAPlXIR0xmBPiPAAAUKQUoRGDNicG69SpY7EotWEYio+P1+nTpzVlyhS7dg4AACA/MtlpjUG7rFNoR8R5AAAA9on18luclxObE4OPPvqoxWsnJyeVKFFCzZs3V5UqVezVLwAAANxhxHkAAABFi02JwfT0dEVERKhNmzYKCQnJqz4BAADgDiPOAwAAKHps2nzExcVFzz//vFJSUvKqPwAAAPmfYccjnyDOAwAA+H+FLM67EZt3JW7YsKG2bduWF30BAACAAxHnAQAAFC02rzEYHR2tQYMG6dixY6pXr56KFStmcb5mzZp26xwAAEB+VFg3HyHOAwAAYPORbD3zzDOaPHmyOnXqJEnq37+/+ZzJZJJhGDKZTMrIyLB/LwEAAPKbAhLs5QZxHgAAwHUKUax3I7lODM6ePVvvvPOODh8+nJf9AQAAwB1GnAcAAFA05ToxaBhXU6Vly5bNs84AAAAUCPZaUDqffBJNnAcAAPAv9oj18kmcdzM2rTFoMpnyqh8AAAAFRmFcY5A4DwAA4CrWGMxBpUqVbho0nj179rY6BAAAgDuPOA8AAKDosSkx+NZbb8nPzy+v+gIAAFAwFLKpxBJxHgAAgBlTibPXuXNnlSxZMq/6AgAAUCAUxqnExHkAAABXFaWpxE65rci6MwAAAIUTcR4AAEDRZPOuxAAAAEVeIZtKTJwHAADwL0wltpaZmZmX/QAAACg4CllikDgPAADgX4pQYjDXU4kBAAAAAAAAFB42bT4CAACAwrn5CAAAAK5i8xEAAADkzLDjYYOIiAiZTCar44UXXpAk9ezZ0+pco0aNbvtxAQAAihQHxHmOwohBAACAAmLTpk3KyMgwv961a5datWqljh07mssefPBBzZw50/zazc3tjvYRAAAABQeJQQAAAFs5aPOREiVKWLx+5513VL58eTVr1sxc5u7urpCQEDt0DgAAoIhi8xEAAADkJGvdGXsckpSYmGhxpKSk3LQPqamp+uqrr/TMM8/IZDKZy1evXq2SJUuqUqVK6tu3r06dOpVXbwMAAEChZM84L78jMQgAAOBg4eHh8vPzMx/jxo276TU//PCDzp8/r549e5rLoqKi9PXXX2vlypV6//33tWnTJt1///25SjQCAACg6GEqMQAAgK3sPJU4NjZWvr6+5mJ3d/ebXjp9+nRFRUUpLCzMXNapUyfzv6tXr6769eurbNmyWrJkiR5//HE7dBgAAKAIKEJTiUkMAgAA2Mhe00Oy2vD19bVIDN7M0aNH9dtvv2nRokU3rBcaGqqyZctq//79t9NNAACAIsUesR5TiQEAAJAnZs6cqZIlS6pt27Y3rJeQkKDY2FiFhobeoZ4BAACgICExCAAAYCvDjoeNMjMzNXPmTPXo0UMuLtcmf1y6dEmDBw/Whg0bdOTIEa1evVrt27dXUFCQHnvssVt+VAAAgCLHQXGeIzCVGAAAwFZ2XmPQFr/99ptiYmL0zDPPWJQ7Oztr586dmjNnjs6fP6/Q0FC1aNFCCxYskI+Pjx06CwAAUESwxiAAAADyo9atW8swrCNNT09PLV++3AE9AgAAQEHFVGIAAAAbmex4AAAAIH9xVJw3ZcoURUZGysPDQ/Xq1dO6detyrPvHH3+oadOmCgwMlKenp6pUqaJJkybZfE9GDAIAANjKgVOJAQAAkMccMJV4wYIFGjBggKZMmaKmTZvqs88+U1RUlPbs2aMyZcpY1S9WrJhefPFF1axZU8WKFdMff/yh5557TsWKFdOzzz6b6/syYhAAAAAAAABwoIkTJ6p3797q06ePqlatqsmTJys8PFxTp07Ntn6dOnXUpUsXVatWTREREerWrZvatGlzw1GG2SExCAAAYCOTYb8DAAAA+Ys947zExESLIyUlxep+qamp2rJli1q3bm1R3rp1a61fvz5Xfd62bZvWr1+vZs2a2fSsJAYBAABsZdjxAAAAQP5ixzgvPDxcfn5+5mPcuHFWtztz5owyMjIUHBxsUR4cHKz4+PgbdrV06dJyd3dX/fr19cILL6hPnz42PSprDAIAAAAAAAB5IDY2Vr6+vubX7u7uOdY1mSy3LDEMw6rseuvWrdOlS5f0559/6rXXXlOFChXUpUuXXPePxCAAAMCtYLQfAABA4WWnWM/X19ciMZidoKAgOTs7W40OPHXqlNUowutFRkZKkmrUqKGTJ09q1KhRNiUGmUoMAABgI9YYBAAAKLzudJzn5uamevXqacWKFRblK1asUJMmTXLdjmEY2a5heCOMGAQAAAAAAAAcaODAgXr66adVv359NW7cWNOmTVNMTIz69esnSRo2bJiOHz+uOXPmSJI++eQTlSlTRlWqVJEk/fHHH5owYYJeeuklm+5LYhAAAMBW9to4hBGDAAAA+Y89Yj0br+/UqZMSEhI0evRoxcXFqXr16lq6dKnKli0rSYqLi1NMTIy5fmZmpoYNG6bDhw/LxcVF5cuX1zvvvKPnnnvOpvuSGAQAALCRvaYBM5UYAAAg/7FHrHcr10dHRys6Ojrbc7NmzbJ4/dJLL9k8OjA7rDEIAAAAAAAAFEGMGAQAALAVU4kBAAAKLwdMJXYUEoMAAAA2YioxAABA4eWoqcSOQGIQhY5nQoZcXDMc3Q04QJvHuzu6C3Awl9BTju4CHCUzVYp3dCcA5LXzUZfl5JXp6G7AAZ7/PPs1t1A0lGp62dFdgAMZ6cnSnz86uhuFFolBAAAAWzGVGAAAoPBiKjEAAAByRGIQAACg8CpCiUF2JQYAAAAAAACKIEYMAgAA2IjNRwAAAAovNh8BAABAzphKDAAAUHgxlRgAAAAAAABAYcaIQQAAABuZDEMm4/Y/BrZHGwAAALAve8R6BSXOIzEIAABgK6YSAwAAFF5MJQYAAAAAAABQmDFiEAAAwEbsSgwAAFB4sSsxAAAAcsZUYgAAgMKLqcQAAAAAAAAACjNGDAIAANiIqcQAAACFF1OJAQAAkDOmEgMAABReTCUGAAAAAAAAUJgxYhAAAMBGTCUGAAAovJhKDAAAgJwxlRgAAKDwYioxAAAAAAAAgMKMEYMAAAC3oKBMDwEAAIDtikqsR2IQAADAVoZx9bBHOwAAAMhf7BHrFZA4j6nEAAAAAAAAQBHEiEEAAAAbsSsxAABA4cWuxAAAAMgZuxIDAAAUXuxKDAAAAAAAAKAwIzEIAABgI1Om/Y7cGjVqlEwmk8UREhJiPm8YhkaNGqWwsDB5enqqefPm2r17dx48PQAAQOF2p+M8RyIxCAAAYCvDjocNqlWrpri4OPOxc+dO87nx48dr4sSJ+vjjj7Vp0yaFhISoVatWunjx4m09KgAAQJHjgDjPUUgMAgAAFBAuLi4KCQkxHyVKlJB0dbTg5MmTNXz4cD3++OOqXr26Zs+ercuXL2vu3LkO7jUAAADyKxKDAAAANsraqc4ehyQlJiZaHCkpKdned//+/QoLC1NkZKQ6d+6sQ4cOSZIOHz6s+Ph4tW7d2lzX3d1dzZo10/r16/P8/QAAAChM7Bnn5XckBgEAAGxlGPY7JIWHh8vPz898jBs3zuqWDRs21Jw5c7R8+XJ9/vnnio+PV5MmTZSQkKD4+HhJUnBwsMU1wcHB5nMAAADIJTvGefmdi6M7AAAAUNTFxsbK19fX/Nrd3d2qTlRUlPnfNWrUUOPGjVW+fHnNnj1bjRo1kiSZTCaLawzDsCoDAAAAsjBiEAAAwEb2nkrs6+trcWSXGLxesWLFVKNGDe3fv9+8O/H1owNPnTplNYoQAAAAN8ZUYgAAAOTMHrvUZR23KCUlRX///bdCQ0MVGRmpkJAQrVixwnw+NTVVa9asUZMmTW79JgAAAEWRg+K8KVOmKDIyUh4eHqpXr57WrVuXY91FixapVatWKlGihHx9fdW4cWMtX77c5nuSGAQAACgABg8erDVr1ujw4cPauHGjnnjiCSUmJqpHjx4ymUwaMGCAxo4dq++//167du1Sz5495eXlpa5duzq66wAAALiJBQsWaMCAARo+fLi2bdume++9V1FRUYqJicm2/tq1a9WqVSstXbpUW7ZsUYsWLdS+fXtt27bNpvuyxiAAAICN7DU9xJY2jh07pi5duujMmTMqUaKEGjVqpD///FNly5aVJA0ZMkRXrlxRdHS0zp07p4YNG+rXX3+Vj4/P7XcUAACgCLFHrGfr9RMnTlTv3r3Vp08fSdLkyZO1fPlyTZ06NduN6SZPnmzxeuzYsfrxxx/1888/q06dOrm+L4lBAAAAW9lrpzkb2pg/f/4Nz5tMJo0aNUqjRo26zU4BAAAUcfaI9f7/+sTERItid3d3q/WkU1NTtWXLFr322msW5a1bt9b69etzdbvMzExdvHhRAQEBNnWTqcQAAAAAAABAHggPD5efn5/5yG7035kzZ5SRkWG1aVxwcLDV5nI5ef/995WUlKQnn3zSpv4xYhAAAMBGjphKDAAAgDvDnlOJY2Nj5evray6/frSgxTUmk8VrwzCsyrIzb948jRo1Sj/++KNKlixpUz9JDAIAANjqNncUtmgHAAAA+Ys9Yr3/v97X19ciMZidoKAgOTs7W40OPHXqlNUowustWLBAvXv31rfffquWLVva3E2mEgMAAAAAAAAO4ubmpnr16mnFihUW5StWrFCTJk1yvG7evHnq2bOn5s6dq7Zt297SvRkxCAAAYCOmEgMAABRejtiVeODAgXr66adVv359NW7cWNOmTVNMTIz69esnSRo2bJiOHz+uOXPmSLqaFOzevbs++OADNWrUyDza0NPTU35+frm+L4lBAAAAW2UaVw97tAMAAID8xR6xno3Xd+rUSQkJCRo9erTi4uJUvXp1LV26VGXLlpUkxcXFKSYmxlz/s88+U3p6ul544QW98MIL5vIePXpo1qxZub4viUEAAAAAAADAwaKjoxUdHZ3tueuTfatXr7bLPUkMAgAA2IrNRwAAAAovO24+kt+x+QgAAAAAAABQBDFiEAAAwEYm2WnzkdtvAgAAAHZmj1ivoMR5JAYBAABsZRhXD3u0AwAAgPzFHrFeAYnzmEoMAAAAAAAAFEGMGAQAALCRybDTVOKC8UEyAABAkWKPWK+gxHkkBgEAAGzFrsQAAACFF7sSAwAAAAAAACjMGDEIAABgI5NhyGSHBaXt0QYAAADsyx6xXkGJ80gMAgAA2Crz/w97tAMAAID8xR6xXgGJ85hKDAAAAAAAABRBjBgEAACwEVOJAQAACi+mEgMAACBn7EoMAABQeLErMQAAAAAAAIDCjBGDAAAAtjKMq4c92gEAAED+Yo9Yr4DEeSQGAQAAbGQyrh72aAcAAAD5iz1ivYIS55EYBAqwp9ps13OPbNK3K6vro+8aS5I83dP03CP/0z21jsqvWLLiz/rou1XV9OO6uxzcW9hDuzZ71bbNPgWXTJIkHY3109ff1NTmbaXk7Jypnl236+66xxUafFFJl9207a9QTf+yjs6e83Jwz2Ev1eqcVYfuR1ShaqICS6To7UG19efq4Gzrvvj6bkV1OKZpEyrrx3kRd7ajAIAb8ll+Rr4/n5bL+XSllvbQ2R5hSqlaLNu6HrsvKWT0Iavy4xMrKa2UhyTJNTZZ/t/Ey/3wFbmcTtPZ7qFKbFsiT58Bt65zjV3qVWe7ShS7rANni+uddU219URYtnXrhsZpYJM/FVn8nDxc03Xioo++3XWX5myvlW39qIr7NeHB3/T7oQj1XxKVl4+BW9S+9T/q+MhuBfhf1tFj/po6s4F2/ZN9PBfgf1nP9tisiuUSVCokUT/8UlWfzmpgUadV8wN69YX/Wl3btms3paU558kzoPAgMQgUUFXKntbDTf/WgWMBFuUvdtigOpXiNGZWc8Un+Ojuqsf0Suf/KuGCl/74K8IxnYXdnE7w0oyv6upEnI8kqVWLgxr12mq9MLitTid4qUK5BM39toYOHSkub+9U9Xtms94atkovDWnr4J7DXjw8M3R4n49++6mUhk/YnmO9Rs1PqnL1Czpzyv3Oda4oYSoxgNvgtf68AmbHKaF3mFIqF5PPb2cVPO6wjk+spIwgtxyvOzapsgyva8vEZ/he+3POlJKp9GA3XW7kr+JzTuRp/3F7Hqx4QK/d+1+9vfpebYsL1ZPVd+uz9kv08NedFXfJx6r+5TQXzf2ruvYmBOpKmovqhsVrZIs1upLmqm93W374H+pzUYPv2aDNx0Pv1OPARs2aHFa/Xpv00ecNtXtvSbVttU//Gf6b+rzyiE6f8baq7+qaqQuJHpq3sKYeb7cnx3aTLrvqmZcfsygjKXgbitBUYjYfyYbJZLrh0bNnT3PdxYsXq3nz5vLx8ZGXl5fuvvtuzZo1y3x+x44dcnd3108//WRxj4ULF8rDw0O7du2SJI0aNUq1a9e2qJOYmKjhw4erSpUq8vDwUEhIiFq2bKlFixbJyOEbLCMjQ+PGjVOVKlXk6empgIAANWrUSDNnzjTX6dmzp/lZXF1dVa5cOQ0ePFhJSVdHIB05ciTHZ//zzz8lSbNmzcr2vIeHh0V/4uPj9dJLL6lcuXJyd3dXeHi42rdvr99//91cJyIiQpMnT7Z6luzeE1zl6Z6mN3uu1Piv79PFy5Z/9Fcrd0rLNlbU9v1hij/ro5//W1UHjweqcpkzDuot7Gnj5nBt2lpKx+N8dTzOV7Pm1lFysouqVDqty5fdNOytVlq7PkLHTvjpn30lNOWLu1WpwlmVCEpydNdhJ1vWl9CXUytq/arsP1WWpMASyXp+yN96742aykg33cHeFR2mTPsduLOI84jz8gO/Jad18f7iuvRAoNJKe+hszzClB7rK59eEG16X6eeiDH9X8yGna//Hp1bw0rluYUpq6i+58n9/ftaj9g4t3FNFC/fcpUPniuuddfco7pK3OtXYnW39f86U0NL9FXXwbIBOXPTV4r2V9N+YcNUNi7Oo52TK1PjWv+mTjXfrWKLvnXgU3IIO7fZo2coKWraykmKP++vTWQ10+kwxtW+9N9v6J097a+rMBvptbXklXXbNsV3DkM6d97Q4cOuKUpzHiMFsxMVd+w92wYIFGjFihPbuvfZD6ul59Qfso48+0oABAzR06FBNmTJFbm5u+vHHH9WvXz/t2rVLEyZMUK1atfTmm2/q2WefVdOmTRUYGKhTp06pX79+euutt1S9evVs+3D+/Hndc889unDhgsaMGaO7775bLi4uWrNmjYYMGaL7779f/v7+VteNGjVK06ZN08cff6z69esrMTFRmzdv1rlz5yzqPfjgg5o5c6bS0tK0bt069enTR0lJSZo6daq5zm+//aZq1apZXBcYGGj+t6+vr8X7Il0NtrMcOXJETZs2lb+/v8aPH6+aNWsqLS1Ny5cv1wsvvKB//vknpy8BbuKVTv/Vhl1ltGVvKXWP2mZxbufBYDWteVRL11fWmQteqlMpTuElL+jDvxs7qLfIK05Ombq38VG5e6Tr773ZTxUq5pWmzEwpKSnnIAKFi8lkaNDbO7Xwy0jFHLL+1Bko6ojzriLOc6D0TLkduqILj5S0KE6u5S2PfZdveGno0H0ypRlKK+2hC4+VVHJ1/p8vaFydMnRXydP6Yksdi/L1MeGqHRqfqzaqBJ1WnZB4ffin5XTS5xts1tkrnlq0p6rqXZc0RP7g4pKhiuUStOAHy98PW/4K012VT99W254e6fpyyndycjJ08EiAZs+vrYNHAm9+IYo8EoPZCAkJMf/bz89PJpPJokySYmNjNWjQIA0YMEBjx441lw8aNEhubm7q37+/OnbsqIYNG2rYsGH66aef9MILL2j+/Pl67rnnVLFiRQ0ePDjHPrz++us6cuSI9u3bp7Cwa2tNVKpUSV26dLH6xDbLzz//rOjoaHXs2NFcVquW9doT7u7u5mfq2rWrVq1apR9++MEiYAwMDLR67n/L7n35t+joaJlMJv3vf/9TsWLX1kupVq2annnmmRyvw43dX++gKpU5o2ffeTTb8x9800RDnlqnRePmKj3DpMxMk8Z/fZ92Hsz5a4WCJaLMOU0et0xubhm6kuyi0e82V8wxf6t6rq4ZeqbbVq1aF6nLV3KeloTC5Ymeh5WRYdJP88o4uiuFG1OJCyzivKuI8xzHOTFDpkwpw8/yT7EMP1c5n7+Y7TXpxV105tlSSo30lCndULG15xU85pDiR5RTyl0kBwsSf89kuTgZSrhsuf5zwhVPBXndODH8e685CvC8ImeToSn/q6+Fe65NI64TGqfH7/pHHeZ1vEELcDRfnxQ5OxtWo/nOnfdQcf8rt9xu7HE/TfikqQ7HFJeXZ5oea/u3Jo35Rf0GP6wT8YwevSVMJcbNfPfdd0pLS8s26Hvuuefk7e2tefPmSZKcnZ01e/Zs/fjjj+ratauWL1+uWbNmydk5+/n+mZmZmj9/vp566imLYDGLt7e3XFyyz+mGhIRo5cqVOn3atk8bPD09lZaWZtM1N3L27FktW7ZML7zwgkWwmCW7T8FtlZKSosTERIujsCtZ/JL6d9ygt2e2UGp69t8DT7TYrbsiT+m1qa3V553HNGVRIw3s/F/Vq3z8DvcWeeXYCV9FD2qrl1+L0uJllTT4pf+qTOnzFnWcnTP1+sC1MjlJH09rkH1DKHQqVLmgRzof1aSR1SUxjSxPGXY8kO8Q590YcZ6dXP/ftGFIpuz/704P89ClBwKVWs5LKZWK6WyfUrpSx0d+i29vhBEc5/r//k2SjJv87u6+8FE9ueAJjV59n56u9ZceqrhfkuTlmqp3Wv2ukSub6Xwy00cLAquvvymbQhv8s7+Efl9XXoeOBmjXP8EaM7GZjsf56tGov2+nm0VbEYrzGDF4i/bt2yc/Pz+Fhlov6urm5qZy5cpp37595rKqVatqwIABeuedd/Tuu++qUqVKObZ95swZnTt3TlWqVLG5XxMnTtQTTzyhkJAQVatWTU2aNNEjjzyiqKicd6P63//+p7lz5+qBBx6wKG/SpImcnCxzxxcuXDAHuhcuXJC3t7fVNb/++qsOHDggwzBy/QxDhw7VG2+8YVGWmpqqu+7KeSfdcePG6a233spV+4VFpTJnFOB7RZ+/9r25zMXZUK0KcXqs2W49NKiH+v5fe3ceF1XV/wH8MwMzDDuCrIosKu6iCSJgpqailcsv83FJjZKMJAkFV3IJFdPc0hItEdTcSiyXfPLBBMVMU1LTJLMUl4QHVwZRloH7+4OHyesMOKODCPN5v17zejHnnnvuucvMfDn3nHMHHEPs571x5ExFb6ELfzugWeObGNbrV2Sea1RbVScDUqlM1Hf+zv/lgBbNbmLQK79j+aouACoaBWNjDsLFuRCTZ/Zmb0Ej0qbjbdjalyD5u4PqNBNTAWMmnMPAEZfwVv8XarF2RHUH4zzGeTWpzMYEghQwuaMSpZsoVRq9CKtT3NwClhl3DFw7qml37iugKpdo9A60N7+Pm/eqb9T7+3/zBp6/6QAH8/sYF3AMe843RxNbJRrbFuCzV/6tziuVVLRInIpYhVc2DMcVpa2B94Qeh7LADGVlEtg/1DvQzrYIt/MN16grCBKc+7MhGrlq74VM9CA2DNYQQRBE87DcvXsXW7duhYWFBTIyMjB58uRq1wXE87joqnXr1jhz5gwyMzNx6NAhHDx4EP3790doaCjWrFmjzrd7925YWVlBpVKhtLQUAwcOxIoVK0Rlbd26Fa1atRKlPXj329raGr/88otoeeW8PPruw6RJk0STfQPA8uXLcfDgQe0rAJg2bRomTpyofq9UKuHu7q7T9uqqzN/d8MacwaK0qaMP4HKuHTb9xxdSqQCZaTmEcvFxLy+XQCqtI7crSH8SQGZaBuCfRsFGrkpMntkHBXf5RFpjsn+PG07+LJ5LJu7TTKTtcUPqTt4YMCSJIEBigOEhhiiDnj7GeYzznoipFCXe5lD8ehf3Ov/TWKP49S7u+ek+5E+eXYSyBvx3rq4pLTfB2TxHBLlfxQ8XvNXpQU2uYv8FT53LkUgEyE0qnmxw4bYdBm78l2h5ZODPsJSVYv7BYOTe5XDzZ4VKZYLzFxzwXPsc/Pizhzr9ufbX8NMxQ37HCWjqeQsXLzcwYJnGxRCxXl2J8/hL8ph8fHyQn5+Pa9euaQwDKSkpwYULF9CzZ0912qRJkyCXy3H48GEEBgZi/fr1GD16tNayHR0d0aBBA2RlPV63X6lUCn9/f/j7+2PChAn48ssvMWrUKMTGxsLLywsA0KNHDyQkJEAmk8HNzQ0ymeaDCdzd3dGsWbNqt1PV8ubNm0MikSArKwuDBg16ZJ0bNmyoUZa9vX2165iZmcHMzLgaPe4Xy3ExR3xcioplUBYq1Okn/nDFu68eRXGpCf57ywq+zXMREnAen6Z0qY0qk4G9+foJHPvFDddvWMLcvBTdu2ajfZv/4oO5PSGVlmPGpANo5n0LM+N7QCoV1HOVFNyVQ6XSPqyN6haFuQpu7v/0MnBxuw9vHyUKlDJczzVHQb64h2iZSoLbN+T4+5LmcD96ApxjsF5jnMc4r6blv+wIx0+voKSpOYqbW8Dqh1swvVGKgt4VN3fsNuXA9FYpbrxXMQLE5rvrUDnJUdJYUTHHYMZtWB7NR97EfxoWoCqH/Grx//4WYHK7FPLs+yhXSKFyqb/Hsi5ad9IXH/X+AWfyHHEq1wVD2pyFq1UBtp6peCBQVOAROFkVYnpqRU/f4e3OIKfAChdu2wEAnnPLRWjHU9j0a8UDLErKTPHnLfGNwYLiinP+cDrVvpTdrTF5/CH88ZcDzv7hiJd7/QGnhoXY/Z8WAIC3RmTCwf4ePv70efU63p63AFQ8YMTOpgjenregUknV84yPfO0kss474u8cG1hYlGJQvyw09byFTxMDnvr+1RtGNMcgGwYf0+DBgzF58mQsXrwYixcvFi1btWoVCgsLMXz4cABAamoq1qxZg4yMDPj6+iI+Ph5RUVHo3bu31iEqUqkUQ4cOxYYNGzBr1iyNgLSwsBBmZmZVzj/zsMphGoWFheo0S0vLaoPBJ2Vvb4+QkBB89tlniIyM1Jh/5s6dOwaZf4Y0fbi2J8YOPIYZb6bBxqIYubes8MVOP+zIaPXolemZZ2d7H5Pe/xH2De7j3j0ZLmY3wAdze+KXU25wdryLwM5XAQAJS74TrTdpRm/8+hsfQFMfNG+txEefH1O/fzu64qmh+3a5YensdrVVLaJ6hXFe9RjnPbl7QXa4VaCCXcp/YXJbhRJ3Bf471RNljhU3d0zvqGB684F5IVUCGmzIgcmtUghyKUrdzfDfqZ643/GfHoamt1Rwm3Je/d521w3Y7rqBotaWyJ3V9KntGz3a9+ebwU5RhHc7Z8LRshDnb9ojfNfLyCmwBgA4Wt6Dq9VddX6JREBU0BE0silAWbkUV/JtsPRwAL4606aqTdAz7MBhL9hYFeP1107BvsF9XLpihw/iX0TejYqenfYN7sOpYaFonVUf71L/7dP0Jno+fxG5eZYYHfEaAMDKsgRR7/yEBnb3ce+eHH9etEf0rL4496fj09sxqrPYMPiYmjRpgoULFyImJgYKhQKjRo2CTCbDjh07MH36dERHRyMgIABKpRJjxoxBTEwMunSp6LEVGRmJlJQUjB07Frt27dJafnx8PNLT0xEQEIB58+bBz88PMpkMGRkZmD9/Po4dO6Y14HrttdcQHByMoKAguLi44OLFi5g2bRp8fHz0nsvm5s2byM3NFaXZ2dmpn5QnCILGcgBwcnKCVCrFypUrERQUhM6dOyMuLg7t27eHSqVCamoqEhISHvtOOYm9v+wV0ftbSgt8tIHziNVXS1cGVbnsv9etEPLqqKdYG6oNpzPt8XKnEJ3zc17BGiIAKDdQOfTMYZzHOO9pKAhpiIKQhlqX3RgnHlKoHOgE5UCnastTOcmRvbW9wepHNWvL6bbYcrqt1mWx+3qK3m/6tR02/arfzb+Hy6Bny67/tMSu/2j/3l70WVeNtD5D3qi2vFXrOmPVOj5w0KAMEevVkTiPDYNPYMKECWjatCkWLVqETz75BGVlZWjTpg0SEhLw5ptvAgCioqJga2srmjxZKpUiKSkJvr6+VQ41adCgAY4cOYKPPvoIc+fOxaVLl9CgQQO0a9cOH3/8MWxttU8eGxISgs2bN2P+/PnIz8+Hi4sLevbsidmzZ+t857lSr169NNI2b96MYcOGAaiY60XbnfCcnBy4uLjAy8sLv/zyC+bNm4fo6Gjk5OTA0dERnTp1QkJCgl51ISIiepZwjsH6j3Ee4zwiIjJexjTHoEQQ6khNiR5BqVTC1tYWAa/EwVSmqO3qUC0wzy2q7SpQLTO9lFfbVaBaoiovwb7cz5Gfnw8bG90n79dX5W9Nz45TYWry5L81qrIi7D/xUY3Xm6iuq/zseSdPh9SCcZ4xMj3JB2gYs0YH7z06E9VbKlURDhyZ+1TiJUPGenUlzmOPQSIiIiJ9CTDQw0eevAgiIiIiMjBDxHp1JM5jwyARERGRvvhUYiIiIqL6y4ieSiyt7QoQERERERERERHR08eGQSIiIiJ9lRvwpaP58+fD398f1tbWcHJywqBBg3Du3DlRntDQUEgkEtGr8mm5RERERKSjpxzn1SY2DBIRERHpqfJJdYZ46erAgQOIiIjAkSNHkJqaCpVKhT59+qCwsFCUr2/fvsjJyVG/9uzZY+jdJyIiIqrXnnacV5s4xyARERFRHfD999+L3iclJcHJyQmZmZno1q2bOt3MzAwuLi5Pu3pEREREVAexxyARERGRvionpDbEC4BSqRS9iouLH1mF/Px8AIC9vb0oPT09HU5OTvDx8cHbb7+NvLw8w+8/ERERUX1mwDjvWceGQSIiIiJ9Gbhh0N3dHba2turX/PnzH7F5ARMnTkTXrl3Rtm1bdXq/fv2wceNG7N+/H4sXL8axY8fQs2dPnRoaiYiIiOh/aqlhcOXKlfDy8oJCoUCnTp2QkZFRZd6cnByMGDECLVq0gFQqRVRU1GPtKocSExEREdWyK1euwMbGRv3ezMys2vzvvfcefv31Vxw6dEiUPnToUPXfbdu2hZ+fHzw8PPDdd9/h1VdfNWyliYiIiMhgtm7diqioKKxcuRLBwcFYvXo1+vXrh7Nnz6JJkyYa+YuLi+Ho6IjY2FgsXbr0sbfLHoNERERE+jJwj0EbGxvRq7qGwfHjx2Pnzp1IS0tD48aNq62mq6srPDw8cP78eYPuPhEREVG9Vgs9BpcsWYIxY8YgLCwMrVq1wrJly+Du7o6EhASt+T09PfHJJ59g9OjRsLW1fexdZY9BIiIiIn2VA5AYqBwdCYKA8ePH45tvvkF6ejq8vLweuc7Nmzdx5coVuLq6PkEliYiIiIyMIWK9/8V5SqVSlGxmZqZxE7ikpASZmZmYOnWqKL1Pnz44fPjwE1akeuwxSERERFQHRERE4Msvv8SmTZtgbW2N3Nxc5Obm4v79+wCAu3fvIiYmBj/99BOys7ORnp6O/v37o2HDhvi///u/Wq49ERERkXHSZS7pGzduoKysDM7OzqJ0Z2dn5Obm1mj92GOQiIiISE8SQYDEAE+a06eMymEk3bt3F6UnJSUhNDQUJiYmOH36NNavX487d+7A1dUVPXr0wNatW2Ftbf3EdSUiIiIyFoaI9SrX12cuaYlE3E1REASNNENjwyARERGRvh7zSXNay9E5a/V5zc3NsXfv3ietEREREREZItZ7aC7p6jRs2BAmJiYavQPz8vI0ehEaGocSExERERERERER1RK5XI5OnTohNTVVlJ6amoqgoKAa3TZ7DBIRERHpq1wAJAboMVhugDKIiIiIyLAMEevpGedNnDgRo0aNgp+fHwIDA/H555/j8uXLCA8PBwBMmzYNf//9N9avX69e5+TJkwAq5pq+fv06Tp48CblcjtatW+u8XTYMEhEREemrFoYSExEREdFTYsChxLoaOnQobt68ibi4OOTk5KBt27bYs2cPPDw8AAA5OTm4fPmyaJ2OHTuq/87MzMSmTZvg4eGB7OxsnbfLhkEiIiIiIiIiIqJaNm7cOIwbN07rsuTkZI20R81BrQs2DBIRERHpzUA9BsEeg0RERETPHkPEenUjzmPDIBEREZG+OJSYiIiIqP6qhaHEtYVPJSYiIiIiIiIiIjJC7DFIREREpK9yAQYZHsKnEhMRERE9ewwR69WROI8Ng0RERET6EsorXoYoh4iIiIieLYaI9epInMehxEREREREREREREaIPQaJiIiI9MWHjxARERHVX0b08BE2DBIRERHpi3MMEhEREdVfRjTHIIcSExERERERERERGSH2GCQiIiLSF4cSExEREdVfHEpMRERERFUSYKCGwScvgoiIiIgMzBCxXh2J8ziUmIiIiIiIiIiIyAixxyARERGRvjiUmIiIiKj+4lBiIiIiIqpSeTmAcgOVQ0RERETPFEPEenUkzuNQYiIiIiIiIiIiIiPEHoNERERE+uJQYiIiIqL6i0OJiYiIiKhKbBgkIiIiqr+MqGGQQ4mJiIiIiIiIiIiMEHsMEhEREemrXABggLvA5XXjTjIRERGRUTFErFdH4jw2DBIRERHpSRDKIQhP/qQ5Q5RBRERERIZliFivrsR5HEpMRERERERERERkhNhjkIiIiEhfgmCY4SF1ZFJqIiIiIqNiiFivjsR5bBgkIiIi0pdgoDkG60jASERERGRUDBHr1ZE4j0OJiYiIiIiIiIiIjBB7DBIRERHpq7wckBhgQuk6Mik1ERERkVExRKxXR+I8NgwSERER6YtDiYmIiIjqLw4lJiIiIiIiIiIiovqMPQaJiIiI9CSUl0MwwFBioY4MMSEiIiIyJoaI9epKnMeGQSIiIiJ9cSgxERERUf3FocRERERERERERERUn7HHIBEREZG+ygVAwh6DRERERPWSIWK9OhLnsWGQiIiISF+CAMAA88bUkYCRiIiIyKgYItarI3EehxITEREREREREREZIfYYJCIiItKTUC5AMMBQYqGO3EkmIiIiMiaGiPXqSpzHhkEiIiIifQnlMMxQYgOUQURERESGZYhYr47EeRxKTERERFSHrFy5El5eXlAoFOjUqRMyMjJqu0pEREREZAD6xnkHDhxAp06doFAo4O3tjVWrVum9TTYMEhEREelJKBcM9tLH1q1bERUVhdjYWJw4cQLPP/88+vXrh8uXL9fQnhIREREZn7oQ5128eBEvvfQSnn/+eZw4cQLTp09HZGQkUlJS9NouGwaJiIiI9CWUG+6lhyVLlmDMmDEICwtDq1atsGzZMri7uyMhIaGGdpSIiIjICNWBOG/VqlVo0qQJli1bhlatWiEsLAxvvfUWFi1apNd2Occg1RuVE3uqSotquSZUW1QqnnujV15S2zWgWqL637l/WpM8q1AKGGBTKpQCAJRKpSjdzMwMZmZmorSSkhJkZmZi6tSpovQ+ffrg8OHDT14ZomdY5We7/H5xLdeEaktZMf91NWaM842bSlXx3f80H+ZhiFivpuO8n376CX369BGlhYSEIDExEaWlpZDJZDrVk9+uVG8UFBQAADL3xtdyTYiIqLYUFBTA1ta2xsqXy+VwcXHBodw9BivTysoK7u7uorRZs2Zh9uzZorQbN26grKwMzs7OonRnZ2fk5uYarD5Ez6LKOC/73cW1XBMiqg3na7sC9Eyo6TgPMHysV5NxXm5urtb8KpUKN27cgKurq051ZMMg1Rtubm64cuUKrK2tIZFIars6T51SqYS7uzuuXLkCGxub2q4O1QJeA8bN2M+/IAgoKCiAm5tbjW5HoVDg4sWLKCkxXO9UQRA0frcevov8oIfzalufqL5hnGfc3/HEa8DYGfv5f1pxHmD4WK+m4zxt+bWlV4cNg1RvSKVSNG7cuLarUetsbGyM8seC/sFrwLgZ8/mv6TvIlRQKBRQKxVPZ1oMaNmwIExMTjbvGeXl5GneLieobxnkVjPk7nirwGjBuxnz+n1acB9ROrPc4cZ6Li4vW/KampnBwcNB523z4CBEREVEdIJfL0alTJ6SmporSU1NTERQUVEu1IiIiIqIn9ThxXmBgoEb+//znP/Dz89N5fkGADYNEREREdcbEiROxZs0arF27FllZWZgwYQIuX76M8PDw2q4aERERET2BR8V506ZNw+jRo9X5w8PDcenSJUycOBFZWVlYu3YtEhMTERMTo9d2OZSYqJ4wMzPDrFmzqp2vgOo3XgPGjeffOAwdOhQ3b95EXFwccnJy0LZtW+zZswceHh61XTUiqkH8jideA8aN5984PCrOy8nJweXLl9X5vby8sGfPHkyYMAGfffYZ3NzcsHz5cgwePFiv7UqEp/m8ZyIiIiIiIiIiInomcCgxERERERERERGREWLDIBERERERERERkRFiwyAREREREREREZERYsMgERERERERERGREWLDIJEWoaGhGDRokEZ6eno6JBIJ7ty5o7GsRYsWkMvl+Pvvv0V5q3slJydXmy83N7fKOqakpCAgIAC2trawtrZGmzZtEB0drV6enJwsKsvV1RX/+te/cPHiRXUeT09Prdv96KOPAADZ2dlV1u3IkSPqckpKSrBw4UL4+vrCwsICDRs2RHBwMJKSklBaWvrYx7Q62srbtm0bFAoFFi5cCACYPXu21rq3bNlSvU737t0RFRUlei+RSLBlyxZR2cuWLYOnp6f6/cPHt/KlUCiqrHN1++rp6Ylly5ZppMfHx8PExER9TirzVnddde/evdp8D5b1sAsXLmD48OFwc3ODQqFA48aNMXDgQPzxxx/qPA+WZW1tDT8/P2zfvl29XNfjri1PeHi4qD5paWl46aWX4ODgAAsLC7Ru3RrR0dEanzN9junDHvU5DQ0NVefdvXs3unfvDmtra1hYWMDf3x/Jycnq5adOnYKZmRl27twp2kZKSgoUCgXOnDmjPkYdOnQQ5VEqlYiNjUXLli2hUCjg4uKCXr16Yfv27ajqOWFlZWWYP38+WrZsCXNzc9jb26NLly5ISkpS5wkNDVXvi0wmg7e3N2JiYlBYWAhAt8+5rtd7bm4uxo8fD29vb5iZmcHd3R39+/fHDz/88Mjzou2YEBHVBMZ5jPMY5zHOY5zHOI/+YVrbFSCqDw4dOoSioiIMGTIEycnJiI2NRVBQEHJyctR53n//fSiVStEXua2tLY4ePQoAOHfuHGxsbETlOjk5ad3evn37MGzYMMTHx2PAgAGQSCQ4e/as6EsZAGxsbHDu3DkIgoDff/8d77zzDgYMGICTJ0/CxMQEABAXF4e3335btJ61tbXG9tq0aSNKc3BwAFARLIaEhODUqVOYM2cOgoODYWNjgyNHjmDRokXo2LHjU/kRWLNmDSIiIvDZZ58hLCxMnd6mTRvs27dPlNfUtPqvPoVCgQ8++ACDBw+GTCarMl/l8X2QRCJ5jNpXLSkpCZMnT8batWsxdepUAMCxY8dQVlYGADh8+DAGDx4sun7kcrl6fV3Ob6WSkhL07t0bLVu2xPbt2+Hq6oqrV69iz549yM/P16hX3759cefOHXz88ccYMmQIDh06hMDAQAC6Hfe3334bcXFxojQLCwv136tXr8a4cePwxhtvICUlBZ6enrh8+TLWr1+PxYsXY8mSJdUfPB09+DndunUrZs6cKTqv5ubmAIAVK1YgKioKU6ZMwcqVKyGXy7Fjxw6Eh4fjzJkzWLRoEXx9fTFjxgyMHTsWwcHBcHBwQF5eHsLDw/Hhhx+ibdu2Wutw584ddO3aFfn5+Zg7dy78/f1hamqKAwcOYPLkyejZsyfs7Ow01ps9ezY+//xzfPrpp/Dz84NSqcTx48dx+/ZtUb6+ffuq/4HLyMhAWFgYCgsLkZCQoM5T3eccePT1np2djeDgYNjZ2WHhwoVo3749SktLsXfvXkREROD333+v6hQQET3TGOcxzqvEOI9xHuM8xnn1ERsGiQwgMTERI0aMwAsvvICIiAhMnz4dcrkcLi4u6jzm5uYoLi4WpT3IyclJ6w+CNrt370bXrl0xadIkdZqPj4/GnVWJRKLenqurK2bNmoWRI0fizz//RIsWLQBUBA9V1amSg4NDlXmWLVuGgwcP4vjx4+jYsaM63dvbG0OGDEFJSYlO+/QkFi5ciJkzZ2LTpk0YPHiwaJmpqekj9+9hw4cPx65du/DFF19g3LhxVeZ78PjWhAMHDuD+/fuIi4vD+vXrcfDgQXTr1g2Ojo7qPPb29gCqvn50Ob+Vzp49iwsXLmD//v3w8PAAAHh4eCA4OFgjr52dHVxcXODi4oJVq1Zhy5Yt2Llzpzpg1OW4W1hYVJnn6tWriIyMRGRkJJYuXapO9/T0RLdu3fTueVCdB+tga2ur9bxeuXIF0dHRiIqKQnx8vDo9OjoacrkckZGRGDJkCAICAjBt2jTs3LkTERER2LJlC9555x00b94cMTExVdZh+vTpyM7Oxh9//AE3Nzd1uo+PD4YPH15lD4Vdu3Zh3LhxGDJkiDrN19dXI5+ZmZl6n0aMGIG0tDR8++23ooCxus858Ojrfdy4cZBIJPj5559haWmpTm/Tpg3eeuutKtcjInrWMc5jnFcTGOcxzmOcR88KDiUmekIFBQX4+uuvMXLkSPTu3RuFhYVIT0+v0W26uLjgt99+U3dX11XlHbHKYR+GsHHjRvTq1UsULFaSyWSiH46aMHXqVMyZMwe7d+/WCBYfl42NDaZPn464uDh1N/zakJiYiOHDh0Mmk2H48OFITEys0e05OjpCKpVi27Zt6jvVupDJZDA1NTXodfX111+jpKQEkydP1rpc13+uDGXbtm0oLS3VGvS98847sLKywubNmwEAJiYmWLduHXbs2IERI0Zg7969SE5OVvfeeFh5eTm2bNmC119/XRQsVrKysqqy94OLiwv279+P69ev67U/5ubmBj1ft27dwvfff4+IiAitn/mnfb6IiAyFcR7jvJrCOI9xHsA4j54NbBgkqsLu3bthZWUlevXr108j35YtW9C8eXO0adMGJiYmGDZs2GP9sDdu3Fi0rco7vdqMHz8e/v7+aNeuHTw9PTFs2DCsXbsWxcXFVa5z9epVfPzxx2jcuDF8fHzU6VOmTNHYz4cD3qCgII08lQHF+fPnRXOKVEfXY6qrf//731iwYAF27NiBXr16ac1z+vRpjW0+OASlKuPGjYNCoah2GEN+fr5G2X369Hlk2Q+faysrK1y+fFmUR6lUIiUlBSNHjgQAjBw5Etu2bYNSqXxk+Q/S5fxWatSoEZYvX46ZM2eiQYMG6NmzJ+bMmYMLFy5UWX5xcTHmzp0LpVKJF198UZ2uy3FfuXKlRp5169YBqLiubGxs4OrqqtN+6nJMn8Qff/wBW1tbrfWRy+Xw9vYWzc/TqlUrREVFYfPmzZg9e7boM/ewGzdu4Pbt2zp/jh60ZMkSXL9+HS4uLmjfvj3Cw8Px73//u9p1fv75Z2zatEl0voDqP+dA9df7n3/+CUEQdN4Hbdflg3foiYhqGuO8fzDO045xHuM8gHEewDjPGHAoMVEVevToIep+DQBHjx5V/4BXSkxMFKWNHDlS3QVen7snGRkZojlBqpsfxdLSEt999x3++usvpKWl4ciRI4iOjsYnn3yCn376ST1/R+UXvCAIuHfvHp577jls375dNDfJpEmTRJPuAhWBw4O2bt2KVq1aidIq74oJgqDzfCu6HlNdtW/fHjdu3MDMmTPh7++vdU6VFi1aaEwQXNXcKw8yMzNDXFwc3nvvPbz77rta81hbW+OXX34RpVXera/Ow+cagHoi6UqbNm2Ct7e3erhAhw4d4O3tjS1btmDs2LGP3EYlXc7vgyIiIjB69GikpaXh6NGj+PrrrxEfH4+dO3eid+/e6nzDhw+HiYkJ7t+/D1tbWyxatEgU/Oty3F9//XXExsaK0irnW9LnugJ0O6Y16eH63r17F1u3boWFhQUyMjKqvCNeuS7wePMWtW7dGmfOnEFmZiYOHTqEgwcPon///ggNDcWaNWvU+Sr/WVOpVCgtLcXAgQOxYsUKUVnVfc6B6q93ffdB23W5fPlyHDx4UKf1iYieFOO8fzDOY5zHOK96jPMY59V3bBgkqoKlpSWaNWsmSrt69aro/dmzZ3H06FEcO3YMU6ZMUaeXlZVh8+bNVQYa2nh5eendDbtp06Zo2rQpwsLCEBsbCx8fH2zduhVvvvkmgH++4KVSKZydnbV2/W7YsKHGfj7M3d29yjw+Pj7IysrSqb66HFN9NGrUCCkpKejRowf69u2L77//XiNokMvlj9y/qowcORKLFi3C3LlzRU+qqySVSh+rbG3n+uF/ENauXYvffvtNlF5eXo7ExES9AkZdzu/DrK2tMWDAAAwYMABz585FSEgI5s6dKwoYly5dil69esHGxkbr5Om6HHdbW9tqr6v8/Hzk5OTodDdZl2P6JCrrc+3aNY1hICUlJbhw4QJ69uypTps0aRLkcjkOHz6MwMBArF+/HqNHj9ZatqOjIxo0aKDz5+hhUqkU/v7+8Pf3x4QJE/Dll19i1KhRiI2NhZeXF4B//lmTyWRwc3PTOtl6dZ/zyu1Utbx58+aQSCTIysrS+lTKh2m7LivnUSIiehoY5/2DcR7jPMZ5jPMY5xk3DiUmegKJiYno1q0bTp06hZMnT6pfkydPrvF5Qh7m6ekJCwsL0VwplV/w3t7eNTYHzIgRI7Bv3z6cOHFCY5lKparxuVuaNGmCAwcOIC8vD3369NF7CEZ1pFIp4uPjkZCQgOzsbIOV+yinT5/G8ePHkZ6eLrquDh48iGPHjuk959CTkEgkaNmypcZ5dHFxQbNmzap8ouKTeu211yCXy7Fw4UKtyw05KbUuBg8eDFNTUyxevFhj2apVq1BYWIjhw4cDAFJTU7FmzRokJyfD19cX8fHxiIqKEj0V70FSqRRDhw7Fxo0bce3aNY3lhYWFUKlUOte1devW6vUqVf6z5uHhUe0TGB+Xvb09QkJC8Nlnn2n9zD/t80VEZAiM8xjn1QTGeYzzHsQ4j54F7DFI9JhKS0uxYcMGxMXFaTyaPiwsDAsXLsSpU6e0PjlKm7y8PBQVFYnSHBwctH65z549G/fu3cNLL70EDw8P3LlzB8uXL0dpaanobp8uCgoKkJubK0qzsLCAjY2N+v3Nmzc18tjZ2UGhUCAqKgrfffcdXnzxRcyZMwddu3aFtbU1jh8/jgULFiAxMREdOnTQq076aty4MdLT09GjRw/06dMHe/fuha2tLYCKoPXhukskEjg7O+tU9iuvvIKAgACsXr1aYx1BEDTKBiqGSUilj3/fJTExEZ07d0a3bt00lgUGBiIxMVH0BLfq6HJ+K508eRKzZs3CqFGj0Lp1a8jlchw4cABr164V9ZTQhS7H/d69exp5zMzM0KBBA7i7u2Pp0qV47733oFQqMXr0aHh6euLq1atYv349rKystAZvNaVJkyZYuHAhYmJioFAoMGrUKMhkMuzYsQPTp09HdHQ0AgICoFQqMWbMGMTExKBLly4AgMjISKSkpGDs2LHYtWuX1vLj4+ORnp6OgIAAzJs3D35+fpDJZMjIyMD8+fNx7NgxrT1NXnvtNQQHByMoKAguLi64ePEipk2bBh8fH73nsqnucw48+npfuXIlgoKC0LlzZ8TFxaF9+/ZQqVRITU1FQkLCY98pJyKqDYzzGOcxzqsa4zzGeYzz6hmBiDS88cYbwsCBAzXS09LSBADC7du3hW3btglSqVTIzc3VWka7du2E8ePH61ymttdPP/2ktez9+/cLgwcPFtzd3QW5XC44OzsLffv2FTIyMtR5kpKSBFtb22r308PDQ+t233nnHUEQBOHixYtV1m3z5s3qcoqKioT58+cL7dq1ExQKhWBvby8EBwcLycnJQmlpqc7HVB/ayrt27ZrQokULwd/fX7h9+7Ywa9YsrXU3MzNTr/PCCy8I77//fpXvBUEQDh8+LAAQPDw81GlJSUlVHpucnBytda5uXz08PISlS5cKxcXFgoODg7Bw4UKtZSxevFho2LChUFxcrFOZ1Z3fh12/fl2IjIwU2rZtK1hZWQnW1tZCu3bthEWLFgllZWXqfACEb775RmsZgiDofNy15QkJCRGVlZqaKoSEhAgNGjQQFAqF0LJlSyEmJka4du2azsdUH4/63OzYsUN4/vnnBUtLS0GhUAidOnUS1q5dq17+5ptvCm3btlWfn0rnz58XLCwshHXr1qmPka+vryjPnTt3hKlTpwrNmzdXf6579eolfPPNN0J5ebnW+nz++edCjx49BEdHR0EulwtNmjQRQkNDhezsbHWeqj57lXT5nOt6vV+7dk2IiIgQPDw8BLlcLjRq1EgYMGCAkJaWps5T1XnRdkyIiGoC4zzGeQ9inMc4rxLjPMZ5xkoiCP+bSZKIiIiIiIiIiIiMBucYJCIiIiIiIiIiMkJsGCQiIiIiIiIiIjJCbBgkIiIiIiIiIiIyQmwYJCIiIiIiIiIiMkJsGCQiIiIiIiIiIjJCbBgkIiIiIiIiIiIyQmwYJCIiIiIiIiIiMkJsGCQiIiIiIiIiIjJCbBgkInqKZs+ejQ4dOqjfh4aGYtCgQU+9HtnZ2ZBIJDh58mSVeTw9PbFs2TKdy0xOToadnd0T100ikeDbb7994nKIiIiInibGeY/GOI/o2cOGQSIyeqGhoZBIJJBIJJDJZPD29kZMTAwKCwtrfNuffPIJkpOTdcqrS5BHRERERP9gnEdEVD3T2q4AEdGzoG/fvkhKSkJpaSkyMjIQFhaGwsJCJCQkaOQtLS2FTCYzyHZtbW0NUg4RERERacc4j4ioauwxSEQEwMzMDC4uLnB3d8eIESPw+uuvq4c5VA4LWbt2Lby9vWFmZgZBEJCfn4+xY8fCyckJNjY26NmzJ06dOiUq96OPPoKzszOsra0xZswYFBUViZY/PMSkvLwcCxYsQLNmzWBmZoYmTZpg3rx5AAAvLy8AQMeOHSGRSNC9e3f1eklJSWjVqhUUCgVatmyJlStXirbz888/o2PHjlAoFPDz88OJEyf0PkZLlixBu3btYGlpCXd3d4wbNw53797VyPftt9/Cx8cHCoUCvXv3xpUrV0TLd+3ahU6dOkGhUMDb2xsffvghVCqV3vUhIiIi0gXjvEdjnEdkvNgwSESkhbm5OUpLS9Xv//zzT3z11VdISUlRD/F4+eWXkZubiz179iAzMxPPPfccXnzxRdy6dQsA8NVXX2HWrFmYN28ejh8/DldXV41A7mHTpk3DggULMGPGDJw9exabNm2Cs7MzgIqgDwD27duHnJwcbN++HQDwxRdfIDY2FvPmzUNWVhbi4+MxY8YMrFu3DgBQWFiIV155BS1atEBmZiZmz56NmJgYvY+JVCrF8uXLcebMGaxbtw779+/H5MmTRXnu3buHefPmYd26dfjxxx+hVCoxbNgw9fK9e/di5MiRiIyMxNmzZ7F69WokJyerg2IiIiKimsY4TxPjPCIjJhARGbk33nhDGDhwoPr90aNHBQcHB+Ff//qXIAiCMGvWLEEmkwl5eXnqPD/88INgY2MjFBUVicpq2rSpsHr1akEQBCEwMFAIDw8XLQ8ICBB8fX21blupVApmZmbCF198obWeFy9eFAAIJ06cEKW7u7sLmzZtEqXNmTNHCAwMFARBEFavXi3Y29sLhYWF6uUJCQlay3qQh4eHsHTp0iqXf/XVV4KDg4P6fVJSkgBAOHLkiDotKytLACAcPXpUEARBeP7554X4+HhRORs2bBBcXV3V7wEI33zzTZXbJSIiItIV4zztGOcRUSXOMUhEBGD37t2wsrKCSqVCaWkpBg4ciBUrVqiXe3h4wNHRUf0+MzMTd+/ehYODg6ic+/fv46+//gIAZGVlITw8XLQ8MDAQaWlpWuuQlZWF4uJivPjiizrX+/r167hy5QrGjBmDt99+W52uUqnU89pkZWXB19cXFhYWonroKy0tDfHx8Th79iyUSiVUKhWKiopQWFgIS0tLAICpqSn8/PzU67Rs2RJ2dnbIyspC586dkZmZiWPHjonuHJeVlaGoqAj37t0T1ZGIiIjIEBjnPRrjPCLjxYZBIiIAPXr0QEJCAmQyGdzc3DQmna4MiCqVl5fD1dUV6enpGmXZ2dk9Vh3Mzc31Xqe8vBxAxTCTgIAA0TITExMAgCAIj1WfB126dAkvvfQSwsPDMWfOHNjb2+PQoUMYM2aMaCgOAEgkEo31K9PKy8vx4Ycf4tVXX9XIo1AonrieRERERA9jnFc9xnlExo0Ng0REqAgImzVrpnP+5557Drm5uTA1NYWnp6fWPK1atcKRI0cwevRoddqRI0eqLLN58+YwNzfHDz/8gLCwMI3lcrkcQMWd10rOzs5o1KgRLly4gNdff11rua1bt8aGDRtw//59dVBaXT20OX78OFQqFRYvXgyptGJ62q+++kojn0qlwvHjx9G5c2cAwLlz53Dnzh20bNkSQMVxO3funF7HmoiIiOhJMM6rHuM8IuPGhkEiosfQq1cvBAYGYtCgQViwYAFatGiBa9euYc+ePRg0aBD8/Pzw/vvv44033oCfnx+6du2KjRs34rfffoO3t7fWMhUKBaZMmYLJkydDLpcjODgY169fx2+//YYxY8bAyckJ5ubm+P7779G4cWMoFArY2tpi9uzZiIyMhI2NDfr164fi4mIcP34ct2/fxsSJEzFixAjExsZizJgx+OCDD5CdnY1Fixbptb9NmzaFSqXCihUr0L9/f/z4449YtWqVRj6ZTIbx48dj+fLlkMlkeO+999ClSxd1ADlz5ky88sorcHd3x5AhQyCVSvHrr7/i9OnTmDt3rv4ngoiIiMjAGOcxziMyJnwqMRHRY5BIJNizZw+6deuGt956Cz4+Phg2bBiys7PVT5cbOnQoZs6ciSlTpqBTp064dOkS3n333WrLnTFjBqKjozFz5ky0atUKQ4cORV5eHoCKeV2WL1+O1atXw83NDQMHDgQAhIWFYc2aNUhOTka7du3wwgsvIDk5GV5eXgAAKysr7Nq1C2fPnkXHjh0RGxuLBQsW6LW/HTp0wJIlS7BgwQK0bdsWGzduxPz58zXyWVhYYMqUKRgxYgQCAwNhbm6OLVu2qJeHhIRg9+7dSE1Nhb+/P7p06YIlS5bAw8NDr/oQERER1RTGeYzziIyJRDDEpARERERERERERERUp7DHIBERERERERERkRFiwyAREREREREREZERYsMgERERERERERGREWLDIBERERERERERkRFiwyAREREREREREZERYsMgERERERERERGREWLDIBERERERERERkRFiwyAREREREREREZERYsMgERERERERERGREWLDIBERERERERERkRFiwyAREREREREREZER+n9yZBaK7FcQAQAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 1500x500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm_plot, classification_report_2, metrics = plot_count_and_normalized_confusion_matrix(\n",
    "    y_true_2, y_pred_2, display_labels_2, labels_2, xticks_rotation='horizontal')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "20289ecc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Log metrics\n",
    "for metric_name, metric_value in metrics.items():\n",
    "    wandb.log({metric_name: metric_value})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "09a5abfb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Log the confusion matrix matplotlib figure\n",
    "wandb.log({'confusion_matrix_2': wandb.Image(cm_plot)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "2399e29a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Artifact classification_report_2_ds_4_t_1_trn_250>"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Log the classification report as an artifact\n",
    "classification_report_2 = (pd.DataFrame({k: v for k, v in classification_report_2.items() if k != 'accuracy'})\n",
    "                         .transpose().reset_index())\n",
    "\n",
    "wandb.log({'classification_report_2': wandb.Table(\n",
    "    dataframe=classification_report_2)})\n",
    "\n",
    "classification_report_artifact_2 = wandb.Artifact(\n",
    "      f'classification_report_2_{model_name}', type='classification_report')\n",
    "\n",
    "with classification_report_artifact_2.new_file('classification_report_2.txt', mode='w') as f:\n",
    "    f.write(pprint.pformat(classification_report_2))\n",
    "\n",
    "wandb.run.log_artifact(classification_report_artifact_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "b0aa2d8d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "Waiting for W&B process to finish... <strong style=\"color:green\">(success).</strong>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style>\n",
       "    table.wandb td:nth-child(1) { padding: 0 10px; text-align: left ; width: auto;} td:nth-child(2) {text-align: left ; width: 100%}\n",
       "    .wandb-row { display: flex; flex-direction: row; flex-wrap: wrap; justify-content: flex-start; width: 100% }\n",
       "    .wandb-col { display: flex; flex-direction: column; flex-basis: 100%; flex: 1; padding: 10px; }\n",
       "    </style>\n",
       "<div class=\"wandb-row\"><div class=\"wandb-col\"><h3>Run history:</h3><br/><table class=\"wandb\"><tr><td>accuracy</td><td>█▁</td></tr><tr><td>f1</td><td>█▁</td></tr><tr><td>precision</td><td>█▁</td></tr><tr><td>recall</td><td>█▁</td></tr><tr><td>train_loss</td><td>▄▂▃▃▃█▃▄▃▅▁▂▃▁▂▂</td></tr><tr><td>train_mean_token_accuracy</td><td>▄▇▄▅▅▁▇▅▅▅█▇▅▇▇█</td></tr><tr><td>valid_loss</td><td>█▂▄▅▄▇▃▄▇▂▅▂▅▃▄▁</td></tr><tr><td>valid_mean_token_accuracy</td><td>▂▄▃▂▂▁▄▁▄▇▄▄▄▄▅█</td></tr></table><br/></div><div class=\"wandb-col\"><h3>Run summary:</h3><br/><table class=\"wandb\"><tr><td>accuracy</td><td>0.66194</td></tr><tr><td>f1</td><td>0.66194</td></tr><tr><td>precision</td><td>0.66194</td></tr><tr><td>recall</td><td>0.66194</td></tr><tr><td>train_loss</td><td>0.01673</td></tr><tr><td>train_mean_token_accuracy</td><td>1.0</td></tr><tr><td>valid_loss</td><td>0.00772</td></tr><tr><td>valid_mean_token_accuracy</td><td>1.0</td></tr></table><br/></div></div>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       " View run <strong style=\"color:#cdcd00\">finetune_chetGPT_hatespeech_citi</strong> at: <a href='https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new' target=\"_blank\">https://wandb.ai/digdemlab/chatgpt_annotations_llm_comparison/runs/avmc8new</a><br/>Synced 6 W&B file(s), 4 media file(s), 4 artifact file(s) and 0 other file(s)"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "Find logs at: <code>.\\wandb\\run-20250228_105559-avmc8new\\logs</code>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Mark the end of the run\n",
    "wandb.finish()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "chatGPT",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}
